Assessment of the Influence of “Other-Race Effect” on Visual Attention and Perception of Attractiveness Before and After Rhinoplasty

Key Points

Question

Does the race of a casual observer and/or patient influence perception of rhinoplasty?

Findings

This survey study consisted of 134 observers who participated in eye tracking while viewing images of white and Latin American patients who underwent rhinoplasty, and a separate 134 participants who graded the attractiveness of the patients prerhinoplasty and postrhinoplasty. White participants demonstrated increased same-race visual fixation and reported postoperative attractiveness increases across race, whereas Latin American participants did not exhibit racially associated gaze allocation and only reported significant postoperative attractiveness increase in their own racial group.

Meaning

The race of both patients and observers appears to influence visual attention and perceived attractiveness after rhinoplasty.

This survey study measures how the “other-race effect” influences perception of white and Latin American patients undergoing rhinoplasty by using eye-tracking technology and survey methodology.

Abstract

Importance

The “other-race effect” describes the phenomenon in which individuals demonstrate greatest recognition ability among faces of their own race. Thus, in our multicultural world, it follows that race influences social interactions. However, the association of race with perception of plastic surgery outcomes has not been studied.

Objective

To objectively measure how the other-race effect influences perception of white and Latin American patients undergoing rhinoplasty by using eye-tracking technology and survey methodology.

Design, Setting, and Participants

In the first part of the study, 134 participants viewed 32 paired facial images of white and Latin American patients, either prerhinoplasty or postrhinoplasty, on an eye-tracking system that recorded observer scan paths. In the second part of this study, the same patient images were individually graded by a separate group of 134 participants for degree of racial identification and perceived attractiveness.

Main Outcomes and Measures

The primary outcome was to measure the influence of patient and observer race on perception of rhinoplasty outcomes. For the eye-tracking part, planned hypothesis testing was conducted using an analysis of variance to compare patient race, rhinoplasty status, and attractiveness with respect to visual fixation time.

Results

Of the 134 eye-tracking participants, 68 (51%) were women and the mean (SD) age was 26.4 (7.7) years; of the 134 graders, 64 (48%) were women and the mean (SD) age was 25.0 (6.9) years. Rhinoplasty did not affect racial identity scores among either same-race or other-race evaluators. Visual fixation times for white faces were significantly increased compared with Latin American faces among all casual observer groups (white observers mean change, −20.14 milliseconds; 95% CI, −29.65 to −10.62 milliseconds; P < .001; Asian observers mean change, −39.04 milliseconds; 95% CI, −48.95 to −29.15 milliseconds; P < .001; and African American observers mean change, −20.73 milliseconds; 95% CI, −37.78 to −3.69 milliseconds; P < .02), with the exception of Latin American observers (mean change, −7.8 milliseconds; 95% CI, −29.15 to 14.39 milliseconds; P < .51). With respect to attractiveness, white graders reported a significant postrhinoplasty increase across both races (white patients mean change, 8.07 points; 95% CI, 5.01-11.12 points; P < .001; and Latin American patients mean change, 3.69 points; 95% CI, 0.87-6.49 points; P = .01), whereas Latin American graders only observed a significant attractiveness increase in their own race (Latin American patients mean change, 10.50 points; 95% CI, 1.70-19.32 points; P = .02). Neither perceived attractiveness nor rhinoplasty status influenced fixation times.

Conclusions and Relevance

Both patient and observer race influence visual attention and perception of attractiveness before and after rhinoplasty. These findings underscore the importance of counseling patients that the influence of rhinoplasty, as perceived by the casual observer, may vary by race or ethnicity of the observer group.

Level of Evidence

NA.

Introduction

Over the past decade, rhinoplasty has remained the most common facial plastic procedure in both the United States and Latin America.^1,2,3 The continued popularity of rhinoplasty is not unfounded; research has demonstrated that patients with crooked noses are graded as less attractive and are subject to attentional distraction when compared with their straight-nosed counterparts.^4,5 Following rhinoplasty, patients are perceived as being more attractive, and any disadvantages incurred by a crooked nose are reversed.^4,5,6 However, these previous studies were conducted primarily with white patients undergoing rhinoplasty and white lay observers. Thus, a gap exists in our understanding of rhinoplasty’s influence when the patient and casual observer are not racially concordant.

With the growing accessibility and popularity of cosmetic surgery, the patient pool has increased in diversity; currently, one-quarter of all patients undergoing cosmetic procedures in the United States identify as a race other than white.^7,8,9 Furthermore, patients who are not white report significantly lower satisfaction scores postrhinoplasty and higher rhinoplasty revision rates when compared with white patients.^9,10 These data suggest that there is an unmet need for surgeons who are skilled in addressing the needs and expectations of patients of different ethnicities. As such, it is important to understand how the outcomes of rhinoplasty are perceived both within and across different racial groups.

The “other-race effect” (ORE) is a well-described phenomenon in which individuals are better at discriminating between members of their own racial group compared with others.^11,12,13,14 It is a particularly robust phenomenon that has been shown to develop within the first 3 months of age among children of numerous races.^12,15,16,17 An example of ORE would be an individual stating that all members of a certain race “look the same” to them, resulting in an impaired ability to differentiate individuals outside of their own racial group.^11,18 In addition to diminished recognition ability, it has been shown that individuals will display preferential attention toward racially concordant faces when they are presented with faces of varying race.^15,17,19 Thus, ORE is of particular relevance in facial plastic surgery and could influence surgeons’ ability to accurately provide for patients of all races and, on a larger scale, influence the perceived efficacy of cosmetic procedures within a certain racial group by members of another race.

Currently, the interaction between race and facial plastic surgery has not been explored. We sought to investigate this in Latin American individuals because they are the largest minority group within the United States and accounted for 16.3% of the population in 2010.^20,21,22 Latin American individuals also are the largest minority group in cosmetic surgery and accounted for approximately 10% of patients, including a 173% increase in rhinoplasty, over the past decade.^2,7,8,23 For this study, we selected patients who identified as mestizo, a term used in Latin America to refer to someone of mixed indigenous American and European ancestry.^3,21,24,25

Owing to the increasing accessibility of facial plastic surgery and the growing awareness of specialized racial considerations in such procedures, it is imperative that we understand the interplay between race and cosmetic surgery outcomes. This study had 2 goals. The first goal was to use eye tracking to objectively determine how ORE influences rapid visual decision making for casual observers viewing a cohort of white and Latin American patients undergoing rhinoplasty. The second goal was to determine, using a survey, if ORE contributed to differences in how rhinoplasty outcomes were perceived by casual observers from different races.

Methods

Eye Tracking

Participants

The institutional review board at Johns Hopkins University School of Medicine approved this study. Data collection was time bound; casual observers were recruited over 8 weeks from public areas within the Johns Hopkins University campus.

Individuals self-reporting to have an affective psychiatric condition (schizophrenia, autism, or other related spectrum disorders) were excluded owing to established differences in the way individuals with those disorders perceive faces. Individuals younger than 18 years or who had eye-movement disorders were also excluded.^26,27 The participants were naïve to the purpose of the study and were incentivized to participate using a raffle of nominal value.

Stimulus Material

Preoperative and postoperative images of 32 patients (16 of whom identified as white and 16 of whom identified as mestizo) were obtained with written informed consent from the patients. Both groups contained 4 men and 12 women. Postoperative images were taken at least 6 months after surgery. Images were taken in both profile and frontal view with the patient oriented in the Frankfurt plane. Patients were presented in repose on a blue background.

The images of the white and Latin American patients who underwent rhinoplasty were then divided into groups based on sex and age range (18-25, 26-35, and >36), and randomly paired with an image of a patient of a different race within the same sex and age group. This process was repeated again, creating 2 sets of random pairings (groups A and B). Within each set, image pairs of matched white and Latin American patients were displayed side by side in random left/right orientation. Image pairs were matched with respect to operative status (prerhinoplasty or postrhinoplasty) and view (frontal or profile), such that an observer would view a random assortment of both frontal/profile and preoperative/postoperative pairs. The preoperative and postoperative photographs for the same patient were never displayed to observers as a way to reduce priming bias. Between each face pair, a spacer image of a black dot centrally located on a gray background was introduced. This dot was used to ensure that the observer’s gaze would return to a neutral position between scan paths.²⁸

Procedure

Prior to beginning the study, eye-tracking participants were given a brief digital survey via Qualtrics survey software (Qualtrics, Provo, Utah) to collect demographic information and consent to participate. Eye-tracking participants were randomly assigned to group A or B and were shown the paired, life-sized images of patients’ faces at a conversational viewing distance of 60 cm. Images were presented at random for 400 milliseconds on a 27-in LED LG Cinema D43 monitor (LG Electronics, Englewood Cliffs, New Jersey) at a resolution of 1280 × 1024 pixels. A single visual fixation was defined as a period of 200 milliseconds, in which the x and y coordinates remained fixed within 0.5 degrees, so 400 milliseconds was chosen as the total exposure time.⁴ The spacer image was always presented for 1000 milliseconds. Participants were instructed to casually gaze at the paired patient images and focus on the black dot when the spacer was presented. Prior to initiating the study, participants completed a 9-point calibration task and then introduced to the succession of patient images following a 1000-millisecond exposure to the spacer image.

Visual scan paths were recorded using an iView X RED eye-movement monitoring system (SensoMotoric Instruments, Boston, Massachusetts), which uses a remote infrared camera. The eye tracker is a real-time digital image processer that tracks the center of participants’ pupils using the reflection of the infrared camera and measures eye position in a 2-dimensional grid. Gaze was sampled at a rate of 60 Hz.

Data Analysis

Using BeGaze analysis software (SensoMotoric Instruments, Boston, Massachusetts), 2 predetermined areas of interest were outlined for the white and Latin American faces. Visual fixation time (in milliseconds) within each area of interest was calculated by BeGaze. Demographic information and questions regarding exposure to different races were collected from eye-tracking participants at the time of recruitment using Qualtrics survey software and were associated with participants’ eye-tracking data using Stata/SE, version 12.0 (StataCorp LP).

Perceived Attractiveness

Participants

To examine the effect of attractiveness on gaze preference, independent graders (a separate group from the eye-tracking participants) were recruited to grade the attractiveness of the patients in the images. Graders were recruited using survey postings on public-access websites using the Qualtrics platform. As with the eye-tracking participants, demographics and written informed consent were collected at recruitment, and those self-reporting to have an affective psychiatric condition, were younger than 18 years, or who had eye-movement disorders were excluded.

Procedure

Graders randomly viewed paired frontal and profile views of 20 of the 32 patients, with pairs being either prerhinoplasty or postrhinoplasty. To reduce priming, the prerhinoplasty and postrhinoplasty photographs for the same patient were never shown within the same survey. For each image, graders rated their perceived attractiveness of the patient on a sliding 0 to 100 scale (in 1-point increments), with 0 being least attractive and 100 being most attractive. The graders were also asked how closely they identified each patient with their own racial or ethnic group on a sliding 0 to 100 scale (in 1-point increments), with 0 being strong disagreement and 100 being strong agreement.

Data Analysis

Attractiveness grading was collected using Qualtrics and analyzed in Stata/SE. Data were stratified by respondents’ race and analyzed via unpaired, 2-tailed t tests. Attractiveness scores were merged with their images’ respective eye-tracking data. Planned hypothesis testing was conducted using a 3-way analysis of variance to compare patient race, rhinoplasty status, and attractiveness with respect to visual fixation time.

Results

Eye Tracking

Of the 134 eye-tracking participants who successfully completed the study, 51 (38%) were white and 13 (10%) were Latin American (Table 1). Planned hypothesis testing was conducted using a 3-way analysis of variance to compare patient race, observer race, rhinoplasty status, and attractiveness with respect to visual fixation time (Table 2). Analysis of variance of white observer scan paths revealed increased visual fixation on white faces relative to Latin American faces (−20.14 milliseconds; 95% CI, −29.65 to −10.62 milliseconds; P < .001). When controlling for attractiveness, white observers demonstrated no association between gaze allocation and perceived attractiveness (−0.16 milliseconds; 95% CI, −0.52 to 0.21 milliseconds; P = .40). Additionally, rhinoplasty status had no bearing on visual fixation time (0.58 milliseconds; 95% CI, −9.11 to 10.26 milliseconds; P = .91).

Table 1. Observer Demographics.

Participant Characteristics	Eye-Tracking Survey (n = 134)	Attractiveness Grading Survey (n = 134)
Age, mean (SD), y	26.4 (7.7)	25.0 (6.9)
Sex, No. (%)
Male	66 (49)	67 (50)
Female	68 (51)	64 (48)
Other/prefer not to specify	0	3 (2)
Race/ethnicity, No. (%)
White	51 (38)	99 (74)
Latin American	13 (10)	6 (4)
Asian	50 (37)	13 (10)
African American	17 (13)	2 (2)
Other	3 (2)	14 (10)
Education, No. (%)
Less than high school	1 (1)	4 (3)
High school/GED	3 (2)	15 (11)
Some college	7 (5)	36 (27)
2-year college degree	2 (2)	8 (6)
4-year college degree	89 (66)	52 (39)
Master’s degree	20 (15)	16 (12)
Doctoral degree	12 (9)	3 (2)
Annual household income, $
<25 000	58 (44)	37 (28)
26 000-50 000	19 (15)	32 (24)
51 000-75 000	11 (8)	26 (19)
76 000-100 000	11 (8)	14 (10)
101 000-150 000	8 (6)	14 (10)
151 000-200 000	9 (7)	2 (2)
>201 000	15 (12)	9 (7)

Table 2. Multivariate Regression Analysis Comparing Influence of Patient Race, Perceived Attractiveness, and Rhinoplasty Status on Visual Fixation Times.

Observer Fixation Time vs Image Parameters
Covariate	Observer Race	Regression Coefficient (SE)	P Value	95% CI
Patient race	White	−20.14 (4.85)	<.001	−29.65 to −10.62
	Latin American	−7.38 (11.10)	<.51	−29.15 to 14.39
	Asian	−39.04 (5.05)	<.001	−48.95 to −29.15
	African American	−20.73 (8.69)	<.02	−37.78 to −3.69
Attractiveness	White	−0.16 (0.19)	<.40	−0.52 to 0.21
	Latin American	0.46 (0.42)	<.28	−0.37 to 1.29
	Asian	0.21 (0.20)	<.29	−0.17 to 0.59
	African American	0.15 (0.34)	<.65	−0.51 to 0.82
Rhinoplasty	White	0.58 (4.94)	<.91	−9.10 to 10.26
	Latin American	−18.27 (10.98)	<.10	−39.81 to 3.26
	Asian	−5.56 (5.12)	<.28	−15.59 to 4.48
	African American	1.36 (8.78)	<.89	−15.87 to 18.59

No significant gaze trends were seen among Latin American participants (−7.38 milliseconds; 95% CI, −29.15 to 14.39 milliseconds; P = .51), although we were unable to recruit a comparable number of Latin American participants relative to white participants. The Asian (n = 50 [37%]) and African American (n = 17 [13%]) participants also displayed gaze allocation toward white faces (Asian observers, −39.04 milliseconds; 95% CI, −48.95 to −29.15 milliseconds; P < .001; and African American observers, −20.73 milliseconds; 95% CI, −37.78 to −3.69 milliseconds; P = .02), which was also not associated with rhinoplasty status nor attractiveness.

Racial Identity and Perceived Attractiveness

A total of 134 participants, independent of the participants in the eye-tracking portion of the study, completed the racial identity and perceived attractiveness grading survey (Table 1). White graders (n = 99 [74%]) reported that they racially identified with images of white patients significantly more than the images of Latin American patients (Figure 1) (mean [SD]: 77.94 [23.65] for images of white patients vs 47.24 [28.99] for images of Latin American patients; P < .001), a finding that did not change among either racial group with respect to rhinoplasty status. The 6 (4%) Latin American graders who completed this task identified significantly more with the images of Latin American patients than the images of white patients (mean [SD]: 50.59 [27.12] for images of Latin American patients vs 16.14 [18.88] for images of white patients; P < .001). Asian graders (n = 13 [10%]) identified significantly more with images of Latin American patients than images of white patients (mean [SD]: 23.31 [24.98] for images of Latin American patients vs 8.31 [12.83] for images of white patients; P < .001). As with the white graders, rhinoplasty status had no bearing on identity scores for any racial group.

Figure 1. Identity Scores From 134 Independent Participants.

Graders rated their degree of racial identification with patients on a sliding 0 to 100 scale (in 1-point increments), with 0 being strong disagreement and 100 being strong agreement. Error bars represent 95% CIs.

Unpaired, 2-tailed t tests (Figure 2) revealed that white graders reported a statistically significant difference between preoperative and postoperative white patients (8.07 points; 95% CI, 5.01-11.12 points; P < .001) as well as preoperative and postoperative Latin American patients (3.69 points; 95% CI, 0.87-6.49 points; P = .01). In addition, white graders did not report any statistically significant difference in attractiveness between preoperative white and Latin American patients (−2.29 points; 95% CI, −5.23 to 0.65 points; P = .13). However, there was a significant difference in attractiveness scores among postoperative white and Latin American patients, favoring white patients (−6.67 points; 95% CI, −9.60 to −3.75 points; P < .001).

Figure 2. Attractiveness Scores From 134 Independent Participants.

Graders rated their perceived attractiveness of patients on a sliding 0 to 100 scale (in 1-point increments), with 0 being least attractive and 100 being most attractive. Error bars represent 95% CIs. Statistics are the result of unpaired, 2-tailed t tests.

Latin American graders did not appreciate an attractiveness difference between preoperative and postoperative white patients (5.55 points; 95% CI, −4.62 to 16.89 points; P = .26), though they noted a postoperative attractiveness increase in Latin American patients (10.50 points; 95% CI, 1.70-19.32 points; P = .02). With respect to rhinoplasty, Asian graders did not report a significant attractiveness difference with either white patients (0.43 points; 95% CI, −6.80 to 7.66 points; P = .91) or Latin American patients (7.10 points; 95% CI, −0.17 to 14.30 points; P = .06). Among Latin American and Asian graders, no racial group was graded to be significantly more attractive than the other.

Discussion

These data suggest that ORE influences perception of rhinoplasty outcome. To our knowledge, this is the first study to demonstrate this association in facial plastic surgery. Unfortunately, we were restricted in the conclusions we could draw by the limited participant diversity and were unable to ascertain a bidirectional ORE. However, this is an important first step in appreciating that facial plastic surgery may be perceived differently depending on the race or ethnicity of the observer.

When analyzed by observer race, casual observers of all races—with the exception of Latin American observers—exhibited greater fixation times on white faces relative to Latin American faces. This finding was not influenced by either the subject’s perceived attractiveness or rhinoplasty status. These findings suggest that rapid visual decision making may be most influenced by the race of the subject rather than how attractive they are or how closely the observer identifies with them. This phenomenon is likely multifactorial. One explanation could be that white individuals are more inclined to gaze on their own race owing to ORE.¹⁵ However, this finding was seen across multiple races, suggesting that societal influences, such as the overwhelming presence of white individuals in American media, could also be in play.

However, in the attractiveness-rating portion of this study, no significant attractiveness differences were perceived among preoperative patients. This was seen across all observer races, suggesting that, preoperatively, neither race was regarded as more attractive. This changed when observers viewed the faces postrhinoplasty. When the white observers viewed the images of faces after rhinoplasty, they rated both white and Latin American patients as more attractive. Conversely, the Latin American observers in the survey study only noted an attractiveness change in their own race. While our underpowered Latin American participant cohort could have influenced this finding, these data suggest that the observers’ racial identification may have influenced the perceived outcome of rhinoplasty in both same- and other-race subjects. The Asian observers in this study did not report a significant increase in postrhinoplasty attractiveness in white nor Latin American patients. This may have been because of a small sample size.

These findings in aggregate could be explained, at least in part, by the existence of ORE. The other-race effect could theoretically produce such an association because greater same-race facial discernment ability could increase the perceived outcome of rhinoplasty within the subject’s same-race group—as witnessed by the greater postoperative attractiveness boost white participants witnessed in their own race relative to Latin American participants.

However, there are other important factors that may contribute to these differences, such as surgical technique based on geographic location or nasal anatomy between different races. Different facial plastic surgeons conducted the rhinoplasties between the cohorts in this study; the white patients received surgery in the eastern United States, and the Latin American patients received surgery in Colombia. Although there is a great diversity among the white and Latin American patients, on average, the noses of mestizo individuals have a mesorrhine morphology (characterized by thicker skin, a thicker skin-soft tissue envelope, shorter nasal spine, and lower radix and bridge) compared with the leptorrhine noses of the average white patient (thinner skin, thinner skin-soft tissue envelope, longer nasal spine, and higher radix and bridge).^20,23,29,30 These differences in nasal components present specific challenges to the surgeon because rhinoplasty is not a “one size fits all” procedure.

Interestingly, across all races, observer racial identification scores did not change with respect to rhinoplasty status, which suggests that deformity or surgical intervention did not alter racial perception. Because a major concern in ethnic rhinoplasty is the preservation of ethnic identity, it is reassuring to know, despite the small sample size, that Latin American graders racially identified equally with Latin American patients prerhinoplasty and postrhinoplasty.^20,29

Limitations

There are certain limitations that warrant discussion. First, subsequent studies should offer more precise options for participants’ racial and ethnic identification to better comprehend interracial and intraracial interactions. For example, a person of South Asian descent may not identify with a person of East Asian descent—or vice versa—despite both indicating Asian identity on a demographic survey. In addition, although all of the Latin American–identifying individuals recruited for this study resided within the United States, the identifier Latino that was used in the survey is a racialized ethnic label used primarily within the United States and does not carry the same meaning or usage within other countries. Racial identification within some Latin American communities can be influenced by social factors, and these nuances should be taken into account in future efforts.^24,25,31 Further investigation necessitates the recruitment of a larger and more diverse participant pool, both within the United States and abroad.

In addition to recruiting more racially diverse patients and participants, future studies should also require a more diverse participant cohort with respect to racial exposure, education, income, and age. The majority of participants recruited for the present study were well exposed to other races, and were educated, wealthy, and young (Table 1). Because of this, we were unable to detect any significant association between these demographic elements and gaze duration or attractiveness scores. As early racial exposure is a major modifying factor for the development of ORE, further studies should query for the demographic makeup of the environment in which participants grew up.³² In addition, querying for participants’ strength of racial self-identification and measures of implicit bias (such as the implicit association test) would be useful in confirming the presence of ORE.

Finally, the timescales used between the eye-tracking trials (400 milliseconds per image pair) and online survey (unlimited time per patient) were different. In the eye-tracking portion of this study, the faces were displayed in pairs rather than individually because this design, known as a paired comparison task, has been proven to be more sensitive than serial presentation.^12,33 This is because paired comparison prompts the observer to visually choose between stimuli and reveal intrinsic preferences. As such, increased visual fixation toward white faces exhibited at short time intervals may not be present at longer durations, which merits further investigation. It is possible that although racially based fixation prevailed on a short time frame, an attractiveness or nasal deformity–driven fixation could be seen in a longer interval. Likewise, it is unknown if reflexive, 400-millisecond judgments about racial identification and attractiveness would be similar to those generated under unlimited exposure.

Despite these limitations, we believe this pilot study demonstrates important new information about how the influence of facial plastic surgery procedures may vary by patient and observer race or ethnicity. Future directions include investigating different cross-race relationships, types of facial deformity, and means of surgical intervention. We believe that this study helps set the stage for future inquiries to elucidate the effect that both race and cosmetic surgery have on social interaction.

Conclusions

Both patient and observer race influence visual attention and the perceived outcome of rhinoplasty. This pilot study serves as a starting point for future studies—incorporating larger and more diverse cohorts—to investigate the association of race and cosmetic surgery with social perceptions.