A Cultural Comparison of the Facial Inference Process

The purpose of this study was to compare emotion and personality trait attributions to facial expression between American and Indian samples. Data were collected using Amazon.com’s Mechanical Turk (MTurk). Participants in this study were asked to correctly identify the emotion and make inferences from pictures of three different facial expressions (scowling, frowning, and smiling) of young white females and males in six photographs. Each picture was randomly presented for 10 seconds followed by four randomized questions about the individual in the picture. The first question asked participants to identify the emotion shown from a list of six emotions (anger, disgust, fear, happiness, sadness, surprise). The next three questions consisted of a) condensed sets of the Big Five personality traits, b) the three SelfAssessment Manikin dimensions (SAM), ands) various social perceptions. Smiling facial expressions were hypothesized to be inferred as happy and to have the following positive inferences in both cultures: attractive, not threatening, agreeable, extroverted, and pleasing to look at, positive, conscientious, and open-minded a “Halo Effect.” Scowling facial expressions were hypothesized to have the following attributions: anger, unattractive, threatening, excitable, close-minded, not pleasing to look at, bad, negative, dominant, disagreeable, and unconscientiously a “Horns Effect.” Frowning facial expressions were hypothesized to be perceived as: sad, unattractive, good, submissive, not threatening, not pleasing to look at, positive, and calm anin-between effect. Generally, results showed that both cultures attributed the hypothesized emotional and trait attributions to the six facial expressions for all four questions, except for the Indians on the scowling female facial expression across each of the four questions.


vi LIST OF TABLES
performed to escape from danger, to relieve stress, and to gratify some desire. Darwin, influenced by Lamarck (1809Lamarck ( /1991, thought that the constant repetition of these facial expressions led to them becoming inherited by successive generations. Although the Lamarckian origin hypothesis has been refuted, Darwin further hypothesized that humans have the universal ability to instinctively both pose and recognize certain facial expressions. Darwin gathered worldwide evidence suggesting both universal posing and emotional recognition of human facial expressions (such as the expressions related tojoy, pain, anger, terror, disgust, sulkiness, despair, and suffering). He believed that this evidence provided support for instinctive recognition of a set of core underlying emotions from certain facial expressions. Tompkins (1962), following Darwin, suggested the face is a tool of affect that transmits information about the individual to the world and receives information from the world. Tompkin's research centered around the negative affects of shame, distress, and anger, and his last significant work ended with the relationship between affect and cognition of personality (Tomkins, 1963(Tomkins, , 1991(Tomkins, , 1992. In the 1960s, Paul Ekman, Tompkin's protégé, set out to systematically study the universality of emotions for the first time . His research supports Darwin's universality hypothesis for six core facial emotional expressions (angry, sad, happy, disgusted, surprised, and fearful) (Ekman & Friesen, 1971). Ongoing research has supported Ekman's cross-cultural recognition of the six facial expressions (Ekman, 1972(Ekman, , 1973Ekman & Friesen, 1971, 1986Ekman, Sorenson, & Friesen, 1969;Izard, 1971;Ekman & Heider, 1988;Matsumoto, 1992a). There is additional evidence of a seventh core universal emotion, contempt (Ekman & Friesen, 1986;Matsumoto, 1992b).
Cultural similarities of experiences and reactions were found concerning the seven core emotions (angry, sad, happy, disgusted, surprised, fearful, and contempt) between Japanese and American participants (Scherer, Wallbott, Matsumoto, & Kudoh, 1988). The data used in Matsumoto et al. (1988) was part of a larger study involving emotional antecedents and reactions to the same seven emotions across 27 countries (Wallbott & Scherer, 1986). There was significant cross-cultural participant agreement regarding the core expression and emotion connections (Wallbott & Scherer, 1986). Scherer (2010) later confirmed the cross-cultural agreement of emotional elicitation and differentiation regarding the seven emotions. Cross-cultural recognition of emotions is supported by many studies and methodologies, including the emotional recognition of facial expressions (Matsumoto, 2004;Elfenbein & Ambady, 2002a). However, even though the facial recognition of emotions has shown to be quite accurate, there are many CHAPTER II

Stereotypes
Although these expression and emotion associations create convenient inferential shortcuts, stereotyping may reduce the accuracy of these judgments. The word "stereotype" was used early on by Lippmann (1922) in his book Public Opinion. He referred to stereotypes as general cognitive structures that explain error and biases in our interpretation of the world. The most familiar definition of stereotyping is the global process of attributing characteristics to people based on their nationality, ethnicity, gender, or some other visual cue. For example, Malatesta, Fiore, and Messina (1987) found significant evidence of stereotypes for older female faces. Participants misjudged the emotional displays of older female faces in photographs primarily when the emotions were similarly negative or positive. As an example, anger misjudgments correlated with misattributions of disgust and contempt. From these results, participants could have been confusing these three facial expressions among each other.
Regarding culture, cultural stereotypes were found between Hong Kong Chinese and American participants. The Chinese participants were viewed as more emotionally controlled, less open to others, and less extroverted than American participants (Bond, 1986). Results from a study by Olivola and Todorov (2010) revealed that people frequently associate traits based on stereotypes related to facial expressions. One source behind stereotyping traits is the perception that an ambiguous facial expression may resemble another emotionally connected facial expression (Todorov, 2013;Montepare & Dobish, 2003;Neth & Martinez, 2009;Oosterhof & Todorov, 2009;Said, Sebe, & Todorov, 2009;Zebrowitz, Kikuchi, & Fellous, 2010). For example, if a neutral facial expression is perceived as resembling a familiar emotionally related facial expression, then the neutral expression is judged as having the personality traits of that familiar expression (Said et al., 2009;Todorov, 2013). Much stereotyping has been conceptualized within the "Halo Effect" and "Horns Effect."

The Halo and Horns Effects
Evidence of grouping personality traits based on people's appearances was first discovered by Edward Thorndike (1920). He named this phenomenon the "Halo Effect." This occurs when we unconsciously attribute personality traits based on a positive visible global characteristic, such as attractiveness. Nisbett and Wilson (1977) found evidence of a reverse "Halo Effect," commonly referred to as the "Horns Effect." It occurs when a negative visible global characteristic, such as unattractive or threatening, is used to assess other personality traits a person might possess. As a stereotype, the "Halo Effect" is probably a more powerful phenomenon than one might expect. A specific, pleasing visible attribute might actually undermine an individual's ability to accurately assess a person's personality on a global scale (Dion, Berscheid, & Walster, 1972;Nisbett & Wilson, 1977). If we perceive a person as physically attractive, we may assume that other attributes, that we actually know little about, are positive and agreeable as well (Nisbett & Wilson, 1977). The opposite can be assumed about less attractive people.
Attractiveness as a halo and horns effect initiator. The positive associations more attractive faces create in the brain result in perceivers treating attractive individuals with higher respect and receiving them with positivity (Dion et al., 1972). This positive or negative social treatment of individuals based on level of attractiveness may shape personalities and affect confidence, self-perceptions, and behavior in the perceived. The results of the Dion et al. (1972) study support a physical attractiveness stereotype that includes the presumptions that beautiful people will lead happier lives and be more successful. Physical attractiveness was found to affect the social interaction and social influence of the perceived. Attractive people are believed to possess greater material benefits and have greater happiness, which attracts more potential dating partners and marriage prospects in addition to the lure of their appearance. Higher facial attractiveness has also been shown to impact perceptions of positive attributes like trustworthiness.
Furthermore, other traits related to motivation and morality, such as conscientiousness and fidelity are evaluated positively in conjunction with increased attractiveness as a mate selection strategy (Miller, 2007).
Attractiveness is an outward attribute that has evolved in the direction of advertising mate value according to mate selection theorists (Buss & Barnes, 1986;Buss & Schmitt, 1993). Mate value is based on the "good genes" theory, where the quality of an individual's genes is assessed by the inherent presumed personality attributes and behaviors that high and low attractive individuals differentially display (Thornhill & Gangestad, 1993, 1999. Female mate value is judged by attractiveness more than males (Buss & Barnes, 1986;Buss & Schmitt, 1993). Rennels and Kayl (2015) theorized that female attractiveness might be a more accurate indicator of quality, social standing, and/or their behavior compared to men. Especially for unattractive female faces, there is a potential negativity bias for the attractiveness-expressivity association of the perceiver (Principe & Langlois, 2011). Unattractive females elicited more emotional response (disgust) compared to attractive faces (Principe & Langlois, 2011). This may be because less attractive faces create a higher negative visual stimulation than positive or neutral stimuli during early processing (Smith, Cacioppo, Larsen, & Chartrand, 2003). Research has shown that females with low attractiveness are rated more negatively than medium and highly attractive females (Griffin & Langlois, 2006). However, regardless of the rated or rater's gender, more attractive faces take longer to judge than less attractive faces because it is adaptive to prefer healthy (attractive) than unhealthy (unattractive) individuals (Ishai, 2007;Werheid et al., 2007;Zebrowitz, 2004).
Many socially concerned observers once believed that beauty is irrelevant to the trait inference process, but the evidence shows otherwise (Dion, et al., 1972). Physical beauty might be only skin deep, but the effects on our perceptions are unconsciously profound, despite some observers hoping to prove the opposite. Another facial feature that has been examined as an initiator of the halo and horns effects is babyfaced-ness.
Babyfaced-ness and mature faces as halo and horns effects initiators. Some studies have compared the positive reactions to both attractiveness and babyfaced-ness. Zebrowitz and Franklin (2014) investigated the attractiveness halo effect and babyface stereotype (more positive reactions to babyfaced people) reactions to older and younger neutral expressions. Old adult and young adult participants exhibited an attractiveness halo effect and the babyfaced stereotype for both old and young faces, but stronger attractiveness and babyfaced stereotype reactions were found for faces closer to the participants' age. In small claims court, the more attractive plaintiffs were, the more likely they were to win cases. However, when baby-facedness increased in defendants, they won more cases involving negligent actions and were rewarded larger monetary compensation (Zebrowitz & McDonald, 1991).
As the perception of baby-facedness increases, those individuals are judged as higher in honesty and naivety than those with more mature faces (McArthur & Apatow, 1984;Berry & McArthur, 1985;McArthur & Berry, 1987;Zebrowitz & Montepare, 1990). Zebrowitz and Montepare (1992) studied baby-faced individuals across the life span and found that overgeneralizations are made at each age range about the baby-faced having more childlike traits, such as in social autonomy, physical weakness, and intellect, than mature faced peers independent of attractiveness. Results also indicate that attractiveness did not have an independent main effect but babyfaced women and men were perceived as more warm and honest than less attractive babyfaced people. When babyfaced women are less attractive, the babyface stereotype of being sincere was not supported (Berry, 1991). However, there is conflicting evidence of a babyfaced stereotype and attractiveness interaction. The effect of stereotyped perceptions of babyfaces was low for all three levels of attractiveness in young adult males and females compared to less babyfaced peers (Berry, 1991). Babyfaced-ness was not significantly correlated with attractiveness in young or older adults in the Zebrowitz and Franklin (2014) study. The problem might be because babyfaces have been perceived as less intelligent while mature faces are associated with higher intelligence. More socially dependent personality traits are attributed with babyfaced-ness which departs from the independent stereotypes of attractiveness and intelligence. Another obvious clue that can serve as Halo and Horns Effects initiators are facial expressions.

Facial Expressions
Ecological theory suggests that perceiving and responding to an emotion inferred from a particular expression has developed out of adaptive necessity, such as to quickly avoid an angry person (Zebrowitz & Montepare, 2006). Based on a perceived facial display of emotion, a global personality trait evaluation is made during social interactions that shape how perceivers engage with the perceived (Back & Nestler, 2016). For example, in studies by Marsh, Ambady, and Kleck (2005a) and Seidel, Habel, Kirschner, Gur, and Derntl (2010), when people display positive (happy) emotions, they are approached more often than those showing negative (angry) emotions. People are perceived not only as more attractive when they display a happier facial expression than a neutral one (Reis, Wilson, Monestere, Bernstein, Clark, & Seidl, et al., 1990), but they are also treated more positively (Langlois et al., 2000) and are judged as having positive personality attributes overall (Nisbett & Wilson, 1997). Evidence suggests that these personality judgments are more accurate for emotional extremes (happy or sad facial expressions) compared to neutral facial expressions that are attributed a range of emotions due to the ambiguity of the expression (Malatesta et al., 1987).
Previous research has found evidence of a relationship between expressivity and perceived attractiveness because certain expressions, like attractiveness, may truly reflect phenotypic quality, health, and mate value (Buss & Barnes, 1986;Buss & Schmitt, 1993: Thornhill & Gangestad, 1993, 1999. First impressions of strangers and resulting behavior based on facial expressions is well documented in the social sciences (Ekman, 1992;Thorndike, 1920). As indicated above, certain facial expressions may improve or hinder how others perceive your attractiveness and your overall personality (Thorndike, 1920;Nisbett & Wilson, 1977;Rennels & Kayl, 2015;Golle, Mast, & Lobmaier, 2014). By extension, a person's emotional expression, in addition to the level of facial attractiveness, greatly influences how others perceive them which affects the amount of help, attention, rewards, and cooperation they receive from others (Langlois, Kalankanis, Rubenstein, Larson, Hallam, & Smoot, 2000).
Smiling. Some people are born more physically attractive, but the average person can appear more attractive just by smiling. Hall, Schmidt Mast, and West (2016) found that smiling individuals are judged as more attractive and trustworthy than individuals not smiling. The results of Xu et al. (2012) further supports the association between smiling, increased attractiveness, and increased trustworthiness by measuring across cultures.
Despite cultural differences, participants from China also rated highly attractive individuals as more trustworthy (Xu et al., 2012). In a similar study, Chinese participants rated smiling faces as having more "face value" and having multiple positive personality traits. The effect of smiling faces and increased attractiveness was more evident for male faces (Lau, 1982). A study from Brazil investigated whether a closed smile, upper smile, broad smile, or no smile had an effect on personality perception of male and female pictures that ranged in age from young, middle-age, to old. Smiling had an overall effect on improving ratings of attractiveness and kindness. As the degree of smiling increased to a broad smile, individuals in the pictures were rated as happier (Otta, Folladore, Abrosio, & Hoshino, 1996). Different types of smiles make a difference in the attribution of personality traits such as Duchenne smiles (real smiles that involves facial muscles around the eyes) compared to non-Duchenne smiles (Mehu, Little, & Dunbar, 2007). The type of smile impacted ratings of extroversion and generosity, but the differences for generosity was seen mostly in males. Another facial expression that has shown to influence trait inference is scowling.
Scowling. There are significant differences in personality traits attributed to scowling (angry) facial expressions in comparison to other facial expressions. Using the Big Five Personality traits, angry faces (computer generated male faces) were rated highest on extraversion. The lowest rated traits were conscientiousness, neuroticism, openness, and agreeableness in that order (Tidball, Prabhala, & Gallimore, 2006). Angry facial expressions have also been associated with high dominance and low affiliation (Hess, Blairy, & Kleck, 2000;Knutson, 1996;Montepare & Dobish, 2003). The prominent facial feature portraying anger, lowered eyebrows, elicits impressions of dominance particularly in Western cultures (Keating, Mazur, & Segall, 1981). Angry faces also convey information about potential behaviors that include a tendency to attack with a domineering, hostile, and unfriendly manner (Secord, 1958). Marsh et al. (2005a) proposed that angry faces may have evolved to elicit reactions that powerful, maturefaced adults can command. Faces expressing anger were perceived as more mature than faces expressing fear. Angry faces were rated higher on the mature personality traits of independence, strength, dominance, masculinity, coldness, and shrewdness. Participants rated fearful faces higher on personality traits associated with babyfaced-ness such as, dependence, weakness, submissiveness, femininity, warmth, naïveté, honesty, and youthful. When people display angry expressions, they are perceived as more powerful, having higher status, and a higher salary (Tiedens, 2001).
Frowning. Frowning (sad) facial expressions that were computer generated have been associated with low extraversion, agreeableness, and openness, and ratings were higher on conscientiousness and neuroticism for the sad facial expressions (Tidball et al., 2006). Somewhat contradictory evidence was found in the study by Biel et al. (2012).
Sad facial expressions correlated low to moderately with extraversion out of the Big Five factors. The remaining four factors either negatively correlated very weakly or did not correlate at all with the sad facial expression. In addition, sad faces have been associated with low dominance and moderate affiliation (Hess, Blairy, & Kleck, 2000;Knutson, 1996;Montepare & Dobish, 2003).
Sad facial expressions might be associated with low dominance and moderate affiliation because the expression represents distress cues that trigger an evolved inability to exhibit aggressive behaviors, according to the violence inhibition mechanism (VIM; Blair, 2001;Blair et al., 1997) or a concern mechanism (Nichols, 2001). A distress cue (sad or fearful) displayed by the individual infers emotional distress and relates submission of the expresser instead of aggression. The emotions elicited by the distress cues might be an important part of moral socialization (Blair, 2001). However, contradictory of an aversive based reaction, there is evidence of the distress cues not causing primarily aversive emotional responses (Hess, Blairy, & Kleck, 2000;Marsh, Ambady, & Kleck, 2005b). The same distress cues from sad faces resulted in automatically approaching the sad faces initially, but conscious withdrawal subsequently (Seidel, Habel, Kirschner, Gur, & Derntl, 2010). This response is believed to be caused by previous social experiences that resulted in avoidance.

Culture and Emotion Interpretation
Social identity theory created three main theories that attempt to explain cultural differences in emotion interpretation. Social identity theory was one of the original theories created by Tajfel (1972) on the emotional significance and knowledge of social group belonging. From this original theory, identity theories bloomed, creating various new views on group-culture identity. One of those theories, absolutist theory, emphasizes that basic human nature explains the motivation, actions, or characteristics of all people while cultural differences are ignored (Adamopoulos & Lonner, 1994). Although this approach was popular with early psychologists, the term "absolutism" was not coined until 1992 by Berry, Poortinga, Segall, and Dasen.
The second theory, relational identity theory defines similarities between the self cultural identity of the perceived and perceiver in a group interaction context (Burke, 2006). Studies have shown support for relational identity theory (Elfenbein & Ambady, 2002b) that includes a meta-analysis of cross-cultural studies showing a cultural in-group advantage to emotion recognition (Elfenbein & Ambady, 2002a, 2002bMastumoto, 2002). Mastumoto (1989Mastumoto ( , 1992a and Schimmack (1996)  Universalist theory takes a moderate stance that incorporates ideas from the other two theories. Universalism proposes that there are broad commonalities in human nature but also cultural differences (Adamopoulos & Lonner, 1994). Schwartz (2007) indicated that universalism, in the context of moral inclusiveness across countries, can be differentiated by those who are accepting of all cultures despite differences and those who recognize cultural similarities but choose to value their in-group and reject the outgroup because of differences. Countries that were overall more universally inclusive based on morals, value egalitarianism, and do not value embeddedness (restraint on social behaviors that disrupt group cohesiveness). Inclusive countries were more democratic, religiously heterogeneous, Western European, countries that have ruled their territory over 150 years, and were ex-communist countries. Cultures can possibly influence moral inclusiveness universally as well as the universality of certain emotions.
When comparing the universality of emotions across various cultures (U.S., India, China, Argentina, and Japan), results showed that perceiving emotions accurately is more universal in the domain of emotional intelligence which consists of emotional perception, emotion regulation, and emotion understanding. Emotion regulation (managing emotions) and emotion understanding (understanding emotions, emotional language, and the signals conveyed by emotions) were culturally specific (Shao, Doucet, & Caruso, 2015). Triandis and Bhawuk (1997) found that Indian participants had higher agreement on emotion perception, emotion understanding, and emotion regulation than participants from the other countries. Because India is considered a vertical collectivist culture, accuracy in interpreting negative emotions from facial expression is expected to be lower than their accuracy in interpreting emotions from smiling faces (Triandis & Bhawuk, 1997).
Elfenbein and Ambady (2002a) compared American, Indian, and Japanese response bias in emotion recognition of pictures representing seven emotions (happy, sad, angry, surprise, fear, neutral, and disgust) from each culture. Accuracy on recognizing emotions overall was higher for Americans than Indians, and Indians were more accurate than Japanese. The pictures of American facial expressions had higher accuracy ratings of emotion recognition for all participants than pictures of Indian facial expressions and Japanese had the lowest emotion accuracy. The happy and neutral facial expressions were recognized with the highest accuracy and fear and anger had the lowest recognition accuracy across cultures. As hypothesized, there was a significant in-group advantage in emotion recognition of faces from the same culture. Contrary to their hypothesis, however, participants rated out-group pictures of facial expressions with more positive personality traits than expected, this result was especially significant with pictures of Japanese faces. Aside from problems interpreting and expressing negative emotions, there may be cultural differences in regulating and expressing positive emotions for East Asians (Hui, Fok, & Bond, 2009). East Asians were found to regulate their expressions of positive emotions by considering both positive and negative aspects of expression which prevents jealousy and maintains social relationships (Hui, Fok, & Bond, 2009;Mesquita & Albert, 2007).

Hypotheses
This study compared the three researched initiators of "Halo" and "Horns" effects  perceived attractiveness, perceived babyfaced/mature faced, and the facial expressions of smiling, scowling, and frowning. This comparison was made while also examining the reactions of participants from two countries, India and the United States, to the expressions of young female and male Caucasian models.
This study is based on previous research by Radeke and Stahelski (2015 The fourth hypothesis was that Americans will show significantly higher mean accuracy in attributing the predetermined (correct) emotion and personality traits to the appropriate (accurate) facial expressions than the Indian sample, due to the use of Caucasian faces for the three facial expressions.

Research Design and Overview
The study is a 2 (gender) x 3 (facial expression) x 2 (culture) mixed design. The first independent variable, gender, a within-subjects variable, consisted of photographs of either young female or male faces in their early twenties. The second independent variable, facial expression, also a within-subjects variable, presents either a smiling, scowling, or frowning face to participants. The third independent variable, culture, a between-subjects variable, compared differences in how American and Indian participants interpret facial expressions. There are four dependent variables consisting of the four questions specified below that were asked after each of the six pictures were presented. Table 1 specifies the dependent variables.

Participants
There were 1, 097 primarily white U.S. female and male participants around 18-65 years old from a large variety of careers and educational backgrounds were recruited as the first group of participants using Amazon.com's Mechanical Turk (MTurk) online survey platform. Participants were required to be 18 years and older and from the United States to take part in the survey which is ensured by clicking that they agree to the terms and conditions before they can begin the survey. Compensation for participating was $0.50.
For the second group, 892 Asian Indian female and male participants around 18-65 years old from various careers and education levels were recruited using the MTurk survey platform. Although English is probably not the primary language for most

Materials
Dependent variables. The purpose of this study intends to partially replicate significant findings of trait grouping from a previous study by Radeke and Stahelski (2015). In that study, the results showed a prominent pattern of personality trait grouping based on facial expression (smiling, frowning, and scowling) that were utilized to create the questions and answer choices for the survey in this study. The emotion expression ratings showed that the smiling facial expressions (open and closed mouth smiling) were rated significantly as happy, the scowling facial expression was rated significantly as angry, and the frowning facial expression was rated moderately as sad. The results of the As shown in Table 1 for this study, the first question asks, "as quickly as possible, please choose ONE emotion that best describes the emotion of the individual in the photograph." There were six answer choices to select from (angry, sad, happy, surprise, fearful, and disgust. The smiling facial expressions was expected to be associated with happiness, the scowling facial expression was expected to be associated with anger, and the frowning facial expression was expected to be associated with sadness. The second question, "which of the three following groups of personality traits is the BEST fit for the picture above?" assessed the three Self-Assessment Manikin personality dimensions (excited-calm, subordinate-dominant, and positive-negative).
There were three answer choices and each choice grouped about five personality traits based on previous results from the Radeke and Stahelski (2015) study. For the SAM question, the smiling facial expression was expected to be perceived as positive, neither dominant nor submissive, neither calm nor excitable. The scowling facial expression should be associated with the personality traits: negative, dominant, and excitable. The frowning facial expression was expected to be perceived as negative, submissive, and calm .
The third question, "which of the three following groups of personality traits is the BEST fit for the picture above?" assessed perceived attractiveness, pleasantness, threat, and honesty. Similar to the second question, three answer choices with grouped personality traits were created based on results previously mentioned. Participants were expected to perceive the smiling facial as pleasing to look at, attractive, not threatening, and good. While the scowling facial was expected to be associated with: not pleasing to look at, not attractive, threatening, and bad. The frowning facial expression was expected to be associated with: not pleasing to look at, unattractive, not threatening, and bad.
The fourth question, "which of the three following groups of personality traits is the BEST fit for the picture above?" measured the Big-Five personality traits (agreeableness, openness, conscientiousness, extraversion, and neuroticism) (Goldberg, 1992 The frowning facial expression was expected to be associated with: introverted, conscientious, and emotionally stable.

Emotion.
In the first question, participants were asked to view one of the six facial expressions that were randomly presented and answered a brief manipulation check question. The emotion question was to assess what emotion the individual in the picture is facially displaying (anger, sadness, happiness, fear, disgust, and surprise) as perceived by the participant. Based on Ekman's work pertaining to the six universally interpreted facial expressions, the emotion question attempts to measure cross-cultural similarities or differences in the perception of the six expressions (Ekman, 1999). The emotion question was also asked to see if certain emotions are associated with certain specific groupings of the traits and perceptions listed above.

Self-assessment manikin (SAM).
The second question assessed the three SAM dimensions of excited-calm, subordinate-dominant, and positive-negative. The Self-Assessment Manikin was created by  to assess the three affective temperament dimensions of pleasure (positive-negative), arousal (excited-calm), and dominance (dominant-submissive) ranging from positive to negative associations.
Each dimension was measured using a 9-point Likert scale as shown in Appendix A. The SAM scale was not be used in this study but instead, the study used a condensed set of the SAM adjectives that were significantly associated with happy, sad, and angry facial expressions from the Radeke and Stahelski (2015) study as indicated in Table 1.

Mini-markers (MM).
The third question measures the perception of personality traits. It is based on a condensed subset of 40 validated adjectives assessing the Big-Five personality traits, called the Mini-Markers (MM; Saucier, 1994). The original set of 100 personality adjective markers was created by Goldberg (1992) to assess the Big-Five personality factor structure (agreeableness, conscientiousness, emotional stability, extroversion, and openness). Saucier tested the performance of the 100 adjective markers in 12 sets of data in order to create a validated subset of 40 adjectives that would be simpler and easier to use for certain assessment conditions. As shown in Appendix B, there are eight adjectives representing each of the five factors for the 40 adjective markers.  (2012) MTurk is an inexpensive and convenient tool for recruiting participants from diverse subject pools. Despite concerns over the validity and reliability of MTurk, Berinsky et al. (2012) found MTurk participants to be more representative of the population, inexpensive to recruit, and more consistently responsive to stimuli. They are not an overused pool, and habitual responding was a minor concern. In addition, the sample age range is expected to be notably broader, based on the experiences of Berinksy

Procedure
At least five pilot volunteers were asked to take the survey to calculate the average time it takes to complete the survey. The average time was predicted to be about 12 minutes. Participants who spend more than 30 minutes to complete the survey were excluded from the study because they are most likely not following the instructions that ask participants to answer the four questions for each of the six pictures as quickly as possible. In addition, if the same IP address on MTurk appears more than once, the user ID was checked to determine if the same user has taken the survey multiple times. All of the surveys that the user has completed except for the first one were deleted if they have taken the survey more than once.
Participants were either American or Indian members of Amazon.com who were interested in taking part in the study. They selected the survey (or HIT) from a list of HITS provided by MTurk. As mentioned, a weblink included in the HIT redirected participants to the actual survey in Qualtrics.
After participants are redirected to the survey in Qualtrics, the first question they encountered is their agreement to the terms and conditions of participating in the study, such as the minimum required age and the specific geographic location (either India or the United States). Participants then answered demographic questions after they agreed to take part in the study. Next, participants were asked to view the first picture presented in random order for 10 seconds and then answer four questions that were presented randomly about that picture as quickly as possible. The same procedure was followed for the remaining five pictures. Participants were textually debriefed after finishing the survey and paid within the next few days.

RESULTS
Participants' data were removed if their IP address appeared more than once, if they spent more than 30 minutes responding to the survey, if the country of residence was neither Indian or the United States, and if their survey was about less than 65% complete.
The main effects to be covered in this section will be facial expression, gender, and culture, in that order. The analysis of facial expression is first introduced to simply break down the large effects of this variable by itself and then the small effects of only gender, second, to better understand the complexity of the cultural effects, which incorporates all three variables.
Each Duchenne's smiles when experiencing a pleasant (positive) event or when people were enjoying themselves (Ekman, 1990). Surprise according to Darwin (1872Darwin ( /1989), can be construed as a state of attention that has the potential to graduate into a positive emotion, such as amazement, or shift negative towards fear and terror. Following surprise, the reaction can also morph into sadness that could represent a distress cue triggering a desire to help the expresser (Blair, 2001). However, people expressing sadness have been perceived as likable, warm, and nice (Tiedens, 2001). Although sadness and fear are closely related, fear might have been adaptive to flee from predators or deal with a distressing event that causes a psychological threat (Ekman, 1999). Disgust was listed before anger because a person displaying disgust is normally a result of something offensive whether it is the smell or taste that causes avoidance of the source (Ekman, 1992). As an example, the regional brain activity of participants in Davidson et al.'s (1990) study reflected a desire to withdraw due to the negative affect of a disgusted facial expression (Ekman, 1999). Considered the most negative facial expression of the six, anger elicits impressions of dominance (Keating, Mazur, & Segall, 1981), the potential of attack as a result of a hostile and unfriendly manner (Secord, 1958), and intimidating others to comply (Clark, Pataki, & Carver, 1996;Tiedens, 2001). The remaining three questions all had three answer choices that scaled from the most positive to negative, according to the emotion the answer was associated with based on previous research (Radeke & Stahelski, 2015), and on the emotion question scale.

Facial Expression
Repeated-measures ANOVAs were conducted to solely examine the effects of facial expression across each of the four questions, regardless of culture and gender. A Bonferroni correction was implemented to adjust the  level for each ANOVA to p = 0.0125. For each ANOVA, Mauchly's test was significant, violating the assumption of sphericity,  2 (2) = 648.02, p < .001, therefore the Huynh-Feldt corrected value is reported for the within-subjects effects of each ANOVA. On the emotion question repeatedmeasures ANOVA, as shown in Table 2, within-subjects effects using the Huynh-Feldt correction was significant, F(2, 3109) = 56504.13, p < .001. The facial expression effect size estimate for the emotion question was very high, 2 = 0.97. Participants correctly identified the frowning face as "sad" (M = 3.20, SD = 0.40), the smiling face as "happy" (M = 1.04, SD = 0.26), and the scowling face as "angry" (M = 5.65, SD = 0.55).
For the repeated-measures ANOVA on the SAM question, as shown in Table 3, within-subjects effects using the Huynh-Feldt correction was significant F(2, 3884.9) = 7020, p < .001. The effect size estimate for the SAM question was also high, 2 = 0.78. The repeated-measures ANOVA for the Social Perceptions questions, as shown in Table 4, was significant for within-subjects effects using the Huynh-Feldt correction, F(2, The repeated measures ANOVA for the Big 5 Factors questions, as shown in  Mauchly's test indicated that the assumption of sphericity had been violated, χ 2 (2) = 648.02, p < .001, therefore the Huynh-Feldt corrected value is reported above for the tests of within-subjects effects. The emotion scale is as follows: Happy (Smiling) = 1; Surprise = 2; Sad (Frowning) = 3; Fear = 4; Disgusted = 5; and Angry (Scowling) = 6.  Mauchly's test indicated that the assumption of sphericity had been violated, χ 2 (2) = 54.83, p < .001, therefore the Huynh-Feldt corrected value is reported above for the tests of within-subjects effects. The response scale is as follows: Smiling is Pleasing To Look At, Attractive, Not Threatening, and Good = 1; Frowning is Not Pleasing To Look At, Unattractive, Not Threatening, and Good = 2; and Scowling is Not Pleasing To Look At, Not Attractive, Threatening, and Bad = 3. Table 5 The Big 5 Factors: Facial Expression ANOVA Mauchly's test indicated that the assumption of sphericity had been violated, χ 2 (2) = 2.47, p < .001, therefore the Huynh-Feldt corrected value is reported above for the tests of within-subjects effects. The response scale is as follows: Smiling is Extroverted, Conscientious, Emotionally Stable, and Open-Minded = 1; Frowning is Introverted, Conscientious, and Emotionally Stable = 2; and Scowling is Disagreeable, Unconscientious, Emotionally Stable, and Close-Minded = 3.

Gender
Four additional ANOVAs were performed to analyze gender separately from culture and facial expression across all four questions. A Bonferroni correction was implemented to adjust the  level for each ANOVA to p = 0.0125. For the emotion question repeated-measures ANOVA, as shown in Table 6, within-subjects effects using the Huynh-Feldt correction was significant, F(1,1986) = 108.3, p < .001. However, the effect size estimate was very small, 2 = 0.02. The average answer choices between males and female pictures showed little variability and high agreement for the emotion question.  Table 6 Emotional Inference: Gender ANOVA There are only two conditions, therefore the assumption of sphericity is met. The Greenhouse-Geisser test is reported above for the tests of within-subjects effects. The emotion scale is as follows: Happy (Smiling) = 1; Surprise = 2; Sad (Frowning) = 3; Fear = 4; Disgusted = 5; and Angry (Scowling) = 6.

M SD
The SAM question repeated-measures ANOVA, noted in Table 7, was significant for within-subjects effects using the Huynh-Feldt correction, F(1,1988) = 104.37, p < .001. The effect size estimate was again very small, 2 = 0.05. Male faces compared to female faces showed slightly higher means and, therefore, larger mean differences on the SAM question.
The Social Perceptions question repeated-measures ANOVA, shown in Table 8, showed a significant within-subjects effect using the Huynh-Feldt correction, F(1,1988) = 122.04, p < .001. The effect size estimate was very small, 2 = 0.06. There were larger mean answer choice differences, resulting in low agreement for answer choices between male and female faces.
For the repeated-measures ANOVA regarding the Big 5 Factors, as shown in Table 9, there was a significant within-subjects effects using the Huynh-Feldt correction,  Table 7 Self-Assessment Manikin: Gender ANOVA There are only two conditions, therefore the assumption of sphericity is met. The Greenhouse-Geisser test is reported above for the tests of within-subjects effects. The response scale is as follows: Smiling is Positive, Neither Dominant Nor Submissive, and Neither Calm Nor Excitable = 1; Frowning is Negative, Submissive, and Calm = 2; and Scowling is Negative, Dominant, and Excitable = 3.  Emotion. The first mixed factorial MANOVA, noted in Table 10,

Hypotheses
As proposed in hypothesis 1, that the majority of participants would significantly (accurately) connect smiling to happiness, scowling to anger, and frowning to sadness, with the highest significance ( that Americans had a slightly lower answer choice mean than Indians in attributing the correct predetermined personality traits to the frowning female facial expression but Americans showed slightly higher mean accuracy than Indians for the frowning male facial expression. There was a non-significant mean difference between Americans and Indians for the frowning male facial expression and a significant mean difference for the frowning female facial expression. On the social perceptions question, Americans showed slightly lower accuracy than Indians for the frowning female facial expression and Americans showed slightly higher accuracy with the frowning male facial expression. The same trend was found with a non-significant mean effect of culture for the frowning male facial expression and a significant cultural effect for the frowning female facial expression. For the Big Five Factors question, the pattern continued with Indians showing slightly higher mean accuracy for the frowning female facial expression and Americans showed slightly higher accuracy for the frowning male facial expression.
However, there were non-significant mean differences between Americans and Indians for both the frowning female and male facial expressions. Furthermore, none of these mean differences, significant or non-significant, counter the basic finding  in general most participants made the 'correct' negative responses to the frowning faces.
The fourth hypothesis was not supported, which was that Americans will show significantly higher mean accuracy in attributing the predetermined (correct) emotion and personality traits to the appropriate (accurate) facial expressions than the Indian sample.
Although Americans showed significantly higher mean accuracy for predetermined Evidence of grouping personality traits based on people's appearances was first discovered by Edward Thorndike (1920). He named this phenomenon the "Halo Effect." This occurs when we unconsciously attribute positive personality traits to a person using a global characteristic (such as good, happy, or attractive) (Dion et al., 1972). Clearly there is also a "Horns Effect" which occurs when we use a global characteristic (such as bad, angry, or unattractive) to attribute negative personality traits to a person (Nisbett & Wilson, 1977).
In this study, when sad and angry faces were judged as not attractive, more negative personality traits were associated with individuals. The opposite was true for smiling (happy) faces which were seen as more attractive and attributed with positive personality traits, supporting the Dion et al. (1972) findings. Gender of the model in the photo did not make a difference in trait attribution and attractiveness ratings across each facial expression except for the scowling female facial expression on all four questions.
Indian participants incorrectly attributed disgust to the scowling female facial expression. This inability to see anger in the faces could be because displays of anger are culturally different. People from India may use their whole body to display gesturing and other indicative angry body language while Americans might express anger more discretely and facially. Additionally, body language is based on the situation context in collectivist cultures such as India. In a study by Kapoor et al. (2003), Indians valued interdependent self-construal which aligns with collectivism. The results from Verma and Triandis (1999) found support for the importance of collectivist values to Indians, such as personal relationships and hierarchy. Considering these cultural practices, Indians most likely refrain from displaying extreme emotions, especially anger, with strangers to cultivate valued social relationships and avoid disturbing the harmony of the social group (Matsumoto, 1989(Matsumoto, , 1992aSchimmack, 1996). As a result, this could cause difficulty in identifying facial anger (Matsumoto, 1989(Matsumoto, , 1992aSchimmack, 1996).
Although Indians lean towards vertical collectivism in comparison to Western cultures, there have been results also showing a preference for horizontal collectivism in younger Indians (Verma & Triandis, 1999;Kapoor et al., 2003). Vertical orientation emphasizes hierarchy while horizontal orientation focuses on equality. These presumably

Limitations
The limitations of this study include: a lack of control over the events occurring when participants filled out the survey, the settings where the survey was taken, or how participants filled out the survey. These conditions can influence how the survey was taken and create larger variability in their answers. Some participants spent a longer time than average to finish filling out the survey. If participants took longer than average, then they were potentially not following the directions to view the pictures for 15 seconds and answer the questions quickly, which could invalidate their data. Those who spent a significantly longer amount of time (more than 30 minutes) to complete the survey were removed from the data analysis. Facial structure of the females and males in the photographs could not be specifically controlled to the same degree as the 3D computer generated faces used in Todorov et al. (2013). All faces have slight differences in bone structure that could influence how the face is judged, but computer-generated faces can produce the same facial structures regardless of gender to control for those differences.
Only one set of photographs was used and the models in the photographs were all Caucasian. Using more than one set of photographs could show if similar or different results could be replicated with more than one set to determine the validity of the survey results. For the photographs, Indian participants might have had more difficulty in interpreting the facial expressions because the models in photographs were not Indian, which they would have more familiarity with seeing on a daily basis. Also, English was a second language for most of the Indian participants, which may have led to confusion about the wording in the responses. Finally, there was a lack of previous research support for the scaling used in this study. Despite these limitations, the four MANOVAs were all significant in addition to almost completely correct answer choices across each of the stimulus conditions on all four questions for both cultures.

Future Directions
The results of this study suggest the presence of Halo and Horns effects when encountering a stranger's face for the first time. This is presumably evidence of instantaneously grouping traits based on particular global characteristics. Even though the face is such an important nonverbal communication tool in judging personality traits, it is still unclear which facial factor (age, attractiveness, expression, gender, race, structure) is focused on when people make trait inferences. Two of these factors were investigated in this study. Future research intends to address the remaining four factors, age, attractiveness, race, and facial structure. Additionally, the culture of the perceivers will continue to be studied. Future studies could replicate this study with more countries.
Other future changes will be the separation of answer choices, and Likert scales will be used as the scaling for answer choices.

Summary
The study found support for the universality of the "Halo" and "Horns" effects especially regarding smiling and frowning. The highest accuracy was mixed between the The cultural differences found might be explained by India's collectivist tendencies, in which family, personal connections, and hierarchy are highly valued. India has been referred to as a point where two larger sociocultural areas meet because of the diverging but similar traditions of the northern and the southern regions within India (Dyson & Moore, 1983). Even though India incorporates collectivist and individualist values, Indians orient with vertical collectivism more often (Verma & Triandis, 1999; would strengthen those connections while being sensitive to smaller displays of sadness would also be important to maintain group harmony. Collectivist cultures such as India could be more likely to refrain from extreme negative emotions, especially anger, to preserve social relationships (Matsumoto, 1989(Matsumoto, , 1992aSchimmack, 1996). These could be possible explanations for the Indian participants' difficulty in interpreting the scowling facial expressions despite the high accuracy for the smiling and frowning facial expressions regardless of gender.