Impact of Facial Defect Reconstruction on Attractiveness and Negative Facial Perception

Transcription

1 The Laryngoscope VC 2015 The American Laryngological, Rhinological and Otological Society, Inc. Impact of Facial Defect Reconstruction on Attractiveness and Negative Facial Perception Jacob K. Dey, BS; Masaru Ishii, MD, PhD; Kofi D. O. Boahene, MD; Patrick Byrne, MD; Lisa E. Ishii, MD, MHS Objectives/Hypothesis: Measure the impact of facial defect reconstruction on observer-graded attractiveness and negative facial perception. Study Design: Prospective, randomized, controlled experiment. Methods: One hundred twenty casual observers viewed images of faces with defects of varying sizes and locations before and after reconstruction as well as normal comparison faces. Observers rated attractiveness, defect severity, and how disfiguring, bothersome, and important to repair they considered each face. Results: Facial defects decreased attractiveness (95% confidence interval [CI]: 22.45, 22.08) on a 10-point scale. Mixed effects linear regression showed this attractiveness penalty varied with defect size and location, with large and central defects generating the greatest penalty. Reconstructive surgery increased attractiveness 1.33 (95% CI: 1.18, 1.47), an improvement dependent upon size and location, restoring some defect categories to near normal ranges of attractiveness. Iterated principal factor analysis indicated the disfiguring, important to repair, bothersome, and severity variables were highly correlated and measured a common domain; thus, they were combined to create the disfigured, important to repair, bothersome, severity (DIBS) factor score, representing negative facial perception. The DIBS regression showed defect faces have a 1.5 standard deviation increase in negative perception (DIBS: 1.69, 95% CI: 1.61, 1.77) compared to normal faces, which decreased by a similar magnitude after surgery (DIBS: 21.44, 95% CI: 21.49, 21.38). These findings varied with defect size and location. Conclusions: Surgical reconstruction of facial defects increased attractiveness and decreased negative social facial perception, an impact that varied with defect size and location. These new social perception data add to the evidence base demonstrating the value of high-quality reconstructive surgery. Key Words: Facial deformity, facial defects, facial reconstructive surgery, social perception, attractiveness, mixed effects linear regression. Level of Evidence: NA Laryngoscope, 125: , 2015 INTRODUCTION Facial defects are among the most common facial deformity including scars, lacerations from trauma, hemangiomas, nevi, cutaneous malignancies, and Mohs defects. Further, the human face is a powerful social stimulus that captivates visual attention and provides a wealth of information including identity, emotions, gaze, attraction, intention, and other nonverbal communications that influence social interactions. 1 3 It is therefore important to clearly understand how facial deformities From the Division of Facial Plastic & Reconstructive Surgery (J.K.D., K.D.O.B., P.J.B., L.E.I.), Johns Hopkins School of Medicine, Baltimore, Maryland, U.S.A; and the Division of Rhinology (M.I.), Department of Otolaryngology Head & Neck Surgery, Johns Hopkins School of Medicine, Baltimore, Maryland, U.S.A. Editor s Note: This Manuscript was accepted for publication December 10, The authors have no funding, financial relationships, or conflicts of interest to disclose. Send correspondence to Lisa E. Ishii, MD, MHS, Associate Professor, Division of Facial Plastic & Reconstructive Surgery, Department of Otolaryngology Head & Neck Surgery, Johns Hopkins School of Medicine, 601 N. Caroline Street, Suite 6231, Baltimore, MD learnes2@jhmi.edu DOI: /lary and their surgical repair impact social perception of the face. Attractiveness is an easily observed facial characteristic that can impact the social and psychological outcomes of individuals. The literature shows attractive individuals are more likely to have higher self-esteem, a more positive affect, more fulfilling relationships, achieve greater academic and occupational satisfaction, earn a higher income, and have a better quality of life. 4 7 Therefore, anything that decreases attractiveness and increases negative facial perception could negatively impact the quality of one s life. We previously showed that facial defects decrease attractiveness and increase negative facial perception. 8 We also showed that they negatively impact affect display, the facial display of emotion. 9 These studies contribute to the evidence that facial deformities carry strong social penalties that can incite or exacerbate negative psychosocial outcomes and decrease quality of life We postulate that surgical repair of facial defects will improve social facial perception, and in this study, we quantified how reconstructive surgery impacts attractiveness and negative facial perception. Specifically, we hypothesized that surgical reconstruction of

2 facial defects will increase attractiveness and decrease negative facial perception as graded by the casual observer. We subhypothesized that surgery will restore some categories of facial defects to normal levels of attractiveness. MATERIALS AND METHODS Participants This Johns Hopkins Medicine Institutional Review Board approved study collected complete surveys from 141 casual observers using public access website postings. Twelve were eliminated as outliers, as they rated two or more normal faces as having a defect severity greater than two standard deviations below the mean for defect faces. Nine observers were then randomly eliminated to evenly balance the four surveys with 30 responses each, leaving 120 observers whose data were analyzed. We excluded from recruitment individuals younger than 18 years old and those reporting to have an affective psychiatric condition (autism, schizophrenia, or related spectrum disorders) due to established differences in the way individuals with those disorders direct attention toward a face. 16,17 Observers ranged in age from 18 to 72 years old (mean: years) and were na ıve to the study objectives. Instrument Four image-based surveys were created using photographs of faces with defects before and after surgical reconstruction as well as faces without defects or other obvious deformity, which we refer to as normal faces throughout the article. Our facial plastic surgery image archive was queried using key words to identify facial photographs of patients with facial defects who consented to have their pictures used in research studies. From this pool of images, facial defect patients were randomly selected and characterized into four categories based on defect size and location: small peripheral, small central, large peripheral, large central. Examples of each defect category are presented in Figure 1. Defect size and location were defined as described by Godoy et al. 8 A defect was considered small if its surface area was smaller than the surface area of the patient s iris, approximately 8 mm, the threshold size at which the entire defect can be assessed during a single visual fixation. A defect was considered central if it was located within the central triangle region of the face: including the eyes, nose, and mouth regions. Further, the investigators identified a dichotomy of severity within the large central (LC) group. The more severe LC defects involved subcutaneous tissue, including nasal cartilage, making them more severe and difficult to reconstruct. Therefore, the investigators subdivided the LC category into average/cutaneous-only and difficult/multitissue defects. Facial defect patients continued to be randomly selected from the pool and categorized until each of the four defect groups contained eight patients with pre- and postoperative frontal view photographs in repose. The LC category contained four average and four difficult LC-defect patients. Photographs of 16 normal faces were then selected from our image database to demographically match the defect faces with respect to age, gender, and race. The primary aim was to measure if reconstructive surgery was able to improve social perception; therefore, each defect patient served as his/her own control, and the normal images were used for comparison. In total, there were 16 normal, 32 preoperative defect, and 32 postoperative defect photographs. All postoperative facial defect photographs were more than 6 months out from reconstruction. These photographs were evenly and randomly Fig. 1. Categories of facial defects before and after reconstructive surgery. Example image set of patients with small peripheral (A), small central (B), large peripheral (C), and large central (D) defects before and after reconstructive surgery. [Color figure can be viewed in the online issue, which is available at distributed into four mutually exclusive surveys, such that: 1) each survey contained four photographs per defect category (two preoperative, two postoperative), 2) each survey contained four normal faces, and 3) no survey contained more than one photograph of the same patient, to limit repeated measure effects. Thus, each of the four surveys contained a total of 20 facial photographs. For each image, observers used normalized slider bars with 0.1 increments to rate attractiveness from 0 to 10 (10 5 most attractive), facial defect severity (0 5 no defect, 10 5 most severe), how disfigured and bothersome they found the face (0 5 not disfigured or bothersome, 10 5 extremely disfigured or bothersome), and how important it is to repair the defect (0 5 no defect or not important to repair, 10 5 greatest need for repair). Procedure Observers followed a link on the website posting to randomly complete one of the four surveys. After consenting to 1317

3 TABLE I. Observer-Graded Variables by Defect Group and Operative Status. Attractiveness Severity Disfigured/Bothersome Important to Repair DIBS Defect Group Preop Postop Preop Postop Preop Postop Preop Postop Preop Postop Small peripheral 5.95 (1.93) 6.50 (1.73) 2.37 (2.31) 0.26 (0.67) 2.06 (2.29) 0.21 (0.52) 2.63 (2.91) 0.24 (0.62) (0.74) (0.18) Small central 4.63 (2.03) 6.09 (1.82) 5.22 (2.71) 0.51 (1.30) 5.09 (2.86) 0.39 (1.03) 5.87 (3.02) 0.50 (1.34) (0.87) (0.35) Large peripheral 4.28 (2.14) 5.79 (1.84) 5.96 (2.72) 0.31 (0.91) 5.63 (2.99) 0.24 (0.75) 6.16 (3.10) 0.30 (0.94) 0.16 (0.91) (0.26) Large central: 3.80 (2.18) 5.74 (2.13) 7.50 (1.77) 1.22 (2.17) 7.23 (2.02) 1.10 (2.02) 7.88 (1.84) 1.04 (1.96) 0.67 (0.58) (0.64) average Large central: difficult 2.80 (2.28) 4.58 (2.00) 8.59 (1.47) 3.57 (3.32) 8.25 (1.78) 3.42 (3.29) 8.71 (1.50) 3.56 (3.49) 0.99 (0.49) (1.04) Normal 6.82 (1.64) 0.01 (0.13) 0.01 (0.19) (0.05) (0.05) Data are presented as mean with standard deviation shown in parentheses. The DIBS score has a mean of 0 and a standard deviation of 1. The larger the DIBS score, the more negatively the faces were perceived. DIBS 5 disfigured, important to repair, bothersome, severity factor score; Postop 5 postoperative; Preop 5 preoperative. participate and verifying that the exclusion criteria did not apply, observers entered their demographic information and then completed the 20-image survey. Data Analysis Data were collected using Research Electronic Data Capture (REDCap; and analyzed using Stata 12 SE (Stata Corp., College Station, TX). 18 Analysis of the four observer-rated variables revealed the severity, disfigured/bothersome, and important to repair variables were highly positively correlated with each other and negatively correlated with attractiveness; all correlations were P <.001. Iterated principal factor analysis indicated the severity, disfigured/bothersome, and important to repair variables measured a common domain; thus, they were combined to create the disfiguring, important to repair, bothersome, severity (DIBS) factor score, representing negative facial perception. This normalized factor score accounted for nearly 100% of total variance, and the factor loadings indicated that each variable was represented in the single factor. As shown by Dey et al., the DIBS score is a highly repeatable and valid measure of negative facial perception. 19 Mixed effects linear regression was then used to determine the impact of reconstructive surgery on observer-rated attractiveness and negative facial perception (DIBS score). Mixed effects regression was selected for its superior ability to determine the statistical significance of our findings while accounting for both fixed effects and multiple levels of random effects in the model, such as variance in the facial images (demographic differences, defect differences not related to size or location, differences in reconstructive technique) as well as sources of error inherent in regression modeling. The regression models were created in a bottom-up fashion. The statistically significant variables (P <.05) were first identified and then used to build the respective models using log-likelihood ratio tests to select the models that best fit the data. Last, planned hypothesis testing was conducted using the bias-corrected and accelerated bootstrap method, with 8,000 samples to calculate the mean differences in attractiveness for each postoperative defect category versus normal faces. The experiment-wide a was set at.05. We corrected for multiple comparisons using the Bonferroni method. RESULTS The means and standard deviations of the four observer-graded variables and the calculated DIBS score are presented in Table I for each defect group. These data show the changes in each variable by defect category and with reconstructive surgery. Figure 1 provides example images from each defect category before and after surgery. We first analyzed how surgical reconstruction of facial defects impacts negative facial perception using the DIBS standardized factor score, with a mean of 0 and standard deviation of 1. The preoperative defect faces were used as the base group to create this score. Preoperative defect faces had a 1.5 standard deviation increase in negative perception (DIBS: 1.69, 95% confidence interval [CI]: 1.61, 1.77)] compared to normal faces. Table II presents the DIBS mixed effects regression model, showing the impact of defect size and location as well as reconstructive surgery on the DIBS score. This model also includes interaction terms that account for the interactions between each of the independent variables: size, location, and surgery. Accounting for these interactions generated a model that best fit the data and provided additional information that the magnitude of DIBS decrease after surgery varied depending on preoperative defect size and location. On average, reconstructive surgery decreased DIBS by 1.44 (95% CI: 1.38, 1.50), almost 1.5 standard deviations. Figure 2 is a postregression margins plot that clearly illustrates the key findings. A reference line for normal faces was added for perspective. Figure 2 shows 1) surgery significantly decreased DIBS for all defect categories, 2) the amount of decrease (slope of line) varied based on defect size and location, and 3) surgery decreased the DIBS score to near normal ranges for all defect categories except large central. We next measured facial attractiveness. Preoperatively, faces with defects were rated as less attractive than normal, mean 4.56 (95% CI: 4.41, 4.70) versus 6.82 (95% CI: 6.67, 6.97) on a 10-point scale, respectively. Table III presents the mixed effects linear regression model, quantifying the effects of defect size, defect location, and reconstructive surgery on observer-graded attractiveness. This model also includes interaction terms that account for the interaction of surgery with both defect size and location. The attractiveness penalty varied with defect size and location. On average, large 1318

4 TABLE II. DIBS Regression. Fixed Effects Covariate Regression Coefficient Standard Error P Value 95% Confidence Interval Size < , Location < , Size*location < , Surgery < , Surgery*size < , Surgery*location < , Surgery*size*location < , Constant < , Random Effects Random Effects Parameters Estimate Standard Error 95% Confidence Interval SD: observer , SD: residual , The data represent the mixed effects regression model for observer-rated negative facial perception (DIBS). The fixed effects account for the effect of defect size, location, and surgical status as well as the interactions between those covariates. The constant term represents the DIBS score for all defect faces in the study. The random intercept term (SD: observer) accounts for intrinsic differences in the way observers look at a face. The random error term (SD: residual) accounts for variance in the facial images not accounted for by the fixed effects. Size: 0 5 small, 1 5 large. Location: 0 5 peripheral, 1 5 central. Surgery: 0 5 preoperative, 1 5 postoperative. *Interaction between covariates (e.g., surgery*size: the interaction between surgery and defect size). DIBS 5 disfigured, important to repair, bothersome, severity factor score; SD 5 standard deviation. (21.18, 95% CI: 21.31, 21.03) and central (20.82, 95% CI: 20.96, 20.68) defects incurred the greatest penalty. Surgical reconstruction of facial defects increased attractiveness by an average of 1.33 (95% CI: 1.18, 1.47). The postregression margins plot from this regression is presented in Figure 3, showing the impact of surgery on the respective defect categories. Figure 3 demonstrates 1) surgery significantly increased attractiveness for all defect categories, 2) the amount of attractiveness increase (slope of line) varied with preoperative defect size and location, and 3) surgery increased attractiveness to near normal levels for small defects, especially small peripheral defects. To test the subhypothesis that surgical reconstruction can return faces with defects to the distribution of attractiveness for normal faces, we preformed planned hypothesis tests using the bootstrap method. The results are presented in Table IV. Despite significant improvement in attractiveness with surgery, small but statistically significant differences in attractiveness were measured between all categories of postoperative defect faces and normal faces. Postoperatively, small peripheral defects were closest to normal, only of 10 points less attractive than normal faces. This finding may be due to effect size or sample size limitations within the defect subcategories. With the current study design and sample sizes, we were able to detect attractiveness differences between defect and normal faces of 0.4 with 98% power, 0.3 with 80% power, and 0.2 with 58% power. We conclude that surgical reconstruction of facial defects can improve attractiveness to near normal ranges for small defects. As described in the methods, based on the tissues involved, the LC category was subdivided into average/ cutaneous-only and difficult/multitissue reconstruction. Regression analysis of LC defects demonstrated that within the LC category, defects characterized as difficult were rated 0.5 standard deviations (DIBS: 0.54, 95% CI: 0.43, 0.65) more severe and 1.08 (95% CI: 0.78, 1.38) of Fig. 2. Surgical reconstruction of facial defects decreases negative facial perception. Postregression margins plot illustrating decreased negative facial perception (DIBS) after reconstructive surgery for each defect category. Normal reference line represents the mean DIBS score for normal faces. Error bars represent 95% confidence intervals. DIBS 5 disfigured, important to repair, bothersome, severity factor; Post-op 5 postoperative; Pre-op 5 preoperative. [Color figure can be viewed in the online issue, which is available at

5 TABLE III. Attractiveness Regression. Fixed Effects Covariate Regression Coefficient Standard Error P Value 95% Confidence Interval Size < , Location < , Surgery < , Surgery*size < , Surgery*location < , Constant < , Random Effects Random Effects Parameters Estimate Standard Error 95% Confidence Interval SD: observer , SD: residual , The data represent the mixed effects regression model for observer-rated attractiveness. The fixed effects account for the effect of defect size, location, and surgical status as well as the interaction between surgery and defect size (surgery*size) and the interaction between surgery and defect location (surgery*- location). The constant term represents the attractiveness rating for all defect faces in the study. Size: 0 5 small, 1 5 large. Location: 0 5 peripheral, 1 5 central. Surgery: 0 5 preoperative, 1 5 postoperative. *Interaction between covariates. SD 5 standard deviation. 10 points less attractive preoperatively than the base group of all preoperative LC defects. Furthermore, within the LC category, the difficult defects were not reconstructed to the same optimal level as the average LC defects. These differences were perceived by observers and can be appreciated in the raw data presented in Table I. DISCUSSION We previously developed the methods to measure and analyze the impact of facial defects on observerrated attractiveness at the group/population level, and showed that facial defects penalize attractiveness and social facial perception dependent on defect size and location. 8 The present larger and more powerful study corroborates and expands upon the initial pilot data, and is the first to measure the impact of reconstructing facial defects on attractiveness. The findings of this study contribute new, quantitative evidence that surgical reconstruction of facial defects increases attractiveness and decreases negative facial perception, an improvement based on defect size and location (Tables II and III, Figs. 3 and 4). Among the many findings were major ones that we discuss herein. First, for measures of attractiveness and negative facial perception, there was an ordering based on preoperative defect size and location from least to greatest penalty: small peripheral, small central, large peripheral, large central. Second, reconstructive surgery significantly decreased DIBS and increased attractiveness for all defect categories. Third, the amount of improvement in social perception after surgery depended on preoperative defect size and location, with the most penalized preoperative defects receiving the greatest 1320 magnitude of improvement with surgery. Fourth, surgery decreased DIBS to near normal ranges for all defect categories except large central. Fifth, although surgery increased attractiveness for all categories, only small defects, particularly small peripheral defects, were restored to near normal ranges of attractiveness. These data add to the evidence base demonstrating the value of high-quality reconstructive surgery. We show that facial defects cause significant social penalties that are correctable with reconstructive surgery. In the current healthcare climate, it is important to have Fig. 3. Surgical reconstruction of facial defects increases attractiveness. Postregression margins plot, illustrating increased attractiveness after reconstructive surgery for each defect category. Normal reference line represents the mean attractiveness for normal faces. Error bars represent 95% confidence intervals. Post-op 5 postoperative; Pre-op 5 preoperative. [Color figure can be viewed in the online issue, which is available at www. laryngoscope.com.]

6 TABLE IV. Hypothesis Testing of Attractiveness Ratings Comparing Postoperative to Normal Faces. Comparison Mean Difference 95% Confidence Interval for Difference Postop small peripheral normal , * Postop small central normal , * Postop large peripheral normal , * Postop large central normal , * The data represent the mean difference between the attractiveness ratings for each postoperative (postop) to normal comparison with the adjusted confidence intervals. The confidence intervals were corrected for multiple comparisons and were considered statistically significant (*) if they did not contain 0. quantitative outcomes measures that assess effectiveness and demonstrate measurable benefit to justify intervention. Our outcomes data support the investment in meticulous surgical reconstruction of facial defects. With exception of the LC defect category, faces with defects experienced remarkable improvement and nearnormalization of attractiveness and negative facial perception after reconstructive surgery. Despite significant improvement, the LC defects were still perceived negatively after surgery (Fig. 2). That is, even after surgical reconstruction, the LC defect faces were rated as having a relatively elevated severity, disfigurement, and importance to repair. Subdivision of the LC group into average/ cutaneous-only and difficult/multitissue defects showed that within the LC group, the multitissue defects, in particular, had a suboptimal postoperative outcome compared to the other defect groups (Table I). We highlight these findings as they identify an area for future research and opportunity for surgical innovation to optimize the reconstruction of large central defects, particularly those that involve multiple tissue types. There are a few considerations to be discussed when measuring attractiveness. Observer variability in rating attractiveness is clearly present in our study (standard deviations in Table I). Despite this variation, studies have shown that there is general agreement or consensus when observers rate the attractiveness of faces. 20 However, the variance that does exist in observer ratings emphasizes the importance of using mixed effects linear regression modeling to account for these sources of variation in our analysis. Additionally, we have no evidence that skin cancers or other facial defects occur on people that are as a group less attractive; thus, we assume the measured differences between defect and normal faces are due to the presence of the defect itself. It is important to identify the limitations of this study. The patients included in this study were from a single academic facial plastic surgery division; therefore, these results may not generalize to all facial plastic surgery practices. Also, there are factors that could impact attractiveness ratings not related to defects or reconstructive surgery, such as changes in makeup or hairstyles. Although we do our best to minimize these changes, we cannot eliminate them. However, we can account for them in the residual error of our regression models, allowing us to better and more accurately quantify the effects of the independent variables. In addition, it is important to appreciate that although images were selected at random from our archive and the experiment is performed using random question allocation afforded by modern survey servers, this study suffers from some of the intrinsic biases of classic nonrandom control experiments, that is, an image of a small central defect will always end up in the small central group. This is true for all studies where randomization is not possible (e.g., where normal subjects cannot be randomized to receive defects). These biases must be kept in mind when interpreting our results. CONCLUSION Facial defects induce significant, measurable penalties on social facial perception. Surgical reconstruction of facial defects increased attractiveness and decreased negative facial perception, an impact that varied with defect size and location. These data characterize the benefits of facial reconstructive surgery on social facial perception. BIBLIOGRAPHY 1. Kleinke CL. Gaze and eye contact: a research review. Psychol Bull 1986; 100: Itier RJ, Batty M. Neural bases of eye and gaze processing: the core of social cognition. Neurosci Biobehav Rev 2009;33: Erickson K, Schulkin J. Facial expressions of emotion: a cognitive neuroscience perspective. Brain Cogn 2003;52: Dion K, Walster E, Berschei E. What is beautiful is good. J Pers Soc Psychol 1972;24: Webster M, Driskell JE. Beauty as status. Am J Sociol 1983;89: Umberson D, Hughes M. The impact of physical attractiveness on achievement and psychological well-being. Soc Psychol Q 1987;50: Jackson LA. Physical attractiveness: a sociocultural perspective. In: Cash TF, Pruzinsky T, eds. Body Image: A Handbook of Theory, Research, and Clinical Practice. New York, NY: Guilford Press; 2002: Godoy A, Ishii M, Byrne PJ, Boahene KD, Encarnacion CO, Ishii LE. How facial lesions impact attractiveness and perception: differential effects of size and location. Laryngoscope 2011;121: Godoy A, Ishii M, Dey J, Boahene KD, Byrne PJ, Ishii LE. Facial lesions negatively impact affect display. Otolaryngol Head Neck Surg 2013;149: Macgregor FC. Some psycho-social problems associated with facial deformities. Am Sociol Rev 1951;16: Macgregor FC. Facial disfigurement: problems and management of social interaction and implications for mental health. Aesthetic Plast Surg 1990;14: Levine E, Degutis L, Pruzinsky T, Shin J, Persing JA. Quality of life and facial trauma: psychological and body image effects. Ann Plast Surg 2005;54: Valente SM. Visual disfigurement and depression. Plast Surg Nurs 2009; 29:10 16; quiz Rankin M, Borah GL. Perceived functional impact of abnormal facial appearance. Plast Reconstr Surg 2003;111: ; discussion Brown BC, McKenna SP, Siddhi K, McGrouther DA, Bayat A. The hidden cost of skin scars: quality of life after skin scarring. J Plast Reconstr Aesthet Surg 2008;61: Nakano T, Tanaka K, Endo Y, et al. Atypical gaze patterns in children and adults with autism spectrum disorders dissociated from developmental changes in gaze behaviour. Proc Biol Sci 2010;277: Manor BR, Gordon E, Williams LM, et al. Eye movements reflect impaired face processing in patients with schizophrenia. Biol Psychiatry 1999;46: Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap) a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009;42: Dey JK, Ishii M, Boahene KD, Byrne PJ, Ishii LE. Changing perception: facial reanimation surgery improves attractiveness and decreases negative facial perception. Laryngoscope 2014;124: Berscheid E, Walster E. Physical attractiveness. In: Berkowitz L, ed. Advances in Experimental Psychology. Waltham, MA: Academic Press;