Left wings to the left: Posing and perceived political orientation

ABSTRACT

Images of individuals posing with the left cheek toward the camera are rated as more emotionally expressive than images with the right cheek toward the camera, which is theorized to be due to right hemisphere specialization for emotion processing. Liberals are stereotyped as being more emotional than conservatives. In the present study, we presented images of people displaying either leftward or rightward posing biases in an online task, and asked participants to rate people’s perceived political orientation. Participants rated individuals portrayed with a leftward posing bias as significantly more liberal than those presented with a rightward bias. These findings support the idea that posing direction is related to perceived emotionality of an individual, and that liberals are stereotyped as more emotional than conservatives. Our results differ from those of a previous study, which found conservative politicians are more often portrayed with a leftward posing bias, suggesting differences between posing output for political parties and perceived political orientation. Future research should investigate this effect in other countries, and the effect of posing bias on perceptions of politicians.

KEYWORDS:

• Posing bias
• lateral bias
• political orientation
• right hemisphere
• asymmetry

In photographs and paintings, individuals tend to be portrayed showing more of their left cheek than right (Bruno, Bertamini, & Martinez, 2013; Bruno, Bertamini, Protti, & Guo, 2015; Humphrey & McManus, 1973; LaBar, 1973; Thomas, Burkitt, Patrick, & Elias, 2008b). This posing bias has been found in as disparate contexts as yearbook photos (LaBar, 1973), selfies (Bruno et al., 2013; Bruno et al., 2015), photographs of infants and non-human mammals (Thomas, Burkitt, & Saucier, 2006), and in portraits of Jesus (Acosta, Williamson, & Heilman, 2013). Although the leftward posing bias is widespread, there are certain contexts in which this bias disappears. For example, advertising images show mixed posing biases, where it has been found that women still exhibit a leftward posing bias, but men do not (Thomas, Burkitt, Patrick, & Elias, 2008a), or that posing of both men and women shifts to a rightward bias (Burkitt, Saucier, Thomas, & Ehresman, 2006). The leftward posing bias also disappears in Buddhist artwork (Duerksen, Friedrich, & Elias, 2016) and decreases in images of science academics (Churches et al., 2012). These variances in posing bias tend to be linked to emotionality, where the left cheek is hypothesized to convey more emotion than the right, and thus it has been theorized that posing is one form of communication related to emotional expression (Chatterjee, 2002; Humphrey & McManus, 1973; Nicholls, Clode, Wood, & Wood, 1999). This study explores whether posing bias influences the perceived political orientation of the individual being rated.

This link between posing and emotion communication has a physiological basis. Both experimental and lesion studies show a majority of individuals have a right hemisphere advantage for both perceiving and producing emotions (Bowers, Bauer, Coslett, & Heilman, 1985; Bowers, Blonder, Feinberg, & Heilman, 1991; Dimberg & Petterson, 2000; Hellige, 1993; Killgore & Yurgelun-Todd, 2007; Ley & Bryden, 1979; Safer, 1981; Saxby & Bryden, 1985), and that the left side of the face portrays stronger emotion (Hugdahl, Iversen, & Johnsen, 1993; Moreno, Borod, Welkowitz, & Alpert, 1990; Schiff & Lamon, 1994). This stronger facial expression on the left is thought to be due to the fact that the lower two-thirds of the face are innervated contralaterally, meaning that the left hemiface is innervated by the emotionally dominant right hemisphere (Patten, 1996). Importantly, the left side of the face is also perceived as portraying more intense emotions (Borod, Kent, Koff, Martin, & Alpert, 1988; Indersmitten & Gur, 2003; Sackeim & Gur, 1978). Notably, there is debate over whether this right hemisphere advantage for emotion processing is an oversimplication. Alternative hypotheses, such as the valence hypothesis, which posits that the left hemisphere is dominant for processing positive emotions, and the right hemisphere is dominant for processing negative emotions, have been proposed and received empirical support (for review, see Alves, Fukusima, & Aznar-Casanova, 2008).

In line with this physiological basis of the posing bias, there is evidence that individuals pose with a different directional bias based on how emotional they wish to appear. For example, when individuals were asked to pose for a family portrait, they displayed their left cheek, but this bias was reduced when participants were asked to pose for a scientific or professional portrait (Nicholls et al., 1999). Overall, individuals who report higher emotional expressivity are more likely to pose with their left cheek facing the camera (Nicholls, Clode, Lindell, & Wood, 2002). This shift in bias is further reflected in online profile pictures of academics, where disciplines such as English and psychology, which are more associated with emotion, show a more leftward orientation than science academics (Churches et al., 2012), and in images from different religions, wherein portraits associated with the more emotionally expressive religion Christianity exhibit a leftward posing bias more often than the less emotional Buddhism (Duerksen et al., 2016). Further, images depicting the emotional event of the Crucifixion of Jesus show a greater leftward posing bias than do other portraits (Acosta et al., 2013), and images created with less of an emotional intent, such as advertisements, show a rightward, rather than leftward, bias (Burkitt et al., 2006). While these studies investigate posing biases in disparate contexts, together, they suggest that individuals tend to pose to the left when wishing to express more emotion.

There are also differences in perception of emotion based on pose direction. Individuals posing with a leftward posing bias are judged as more emotional than those posing with a rightward posing bias (Dunstan & Lindell, 2012; Nicholls, Wolfgang, Clode, & Lindell, 2002). Further, portraits of scientists that displayed more of the right cheek were perceived as more scientific (ten Cate, 2002), and participants were more likely to rate individuals posing with a leftward posing bias as English students, and individuals posing with a rightward bias as Chemistry students (Lindell & Savill, 2010). This has led to the suggestion that individuals posing with a leftward posing bias are perceived as more open and creative, and a rightward bias as dry and scientific (Lindell, 2013).

The current study examines perceived political orientation as a function of posing bias because both stereotypes and actual differences in emotionality have been noted based on political group identification. Liberal individuals score higher on empathy on a personality questionnaire (Hirsh, DeYoung, Xiaowen Xu, & Peterson, 2010), as well as lower on emotional stability (Gerber, Huber, Doherty, Dowling, & Ha, 2010). Liberals have also been shown to be more sympathetic toward victims (Robinson, Keltner, Ward, & Ross, 1995) and to endorse compassion and fairness more than conservatives, while conservatives endorse the less emotional values of ingroup loyalty, respect for authorities, and physical and spiritual purity (Graham, Nosek, Haidt, & Young, 2012). Importantly, individuals stereotype liberals and conservatives in line with these characteristics (Graham et al., 2012). Individuals additionally perceive liberals to be more helpful, generous, sympathetic and benevolent, and conservatives to be more disagreeable and unsympathetic (Farwell & Weiner, 2000; Robinson et al., 1995). In the United States, participants described the Republican party (which is conservative) as having traits such as active, independent and decisive, whereas the Democratic party (which is liberal) was described as having traits such as compassionate, devoted to others, emotional, and kind (Winter, 2010). Overall, Republicans are perceived as strong leaders, whereas Democrats are perceived as empathetic and compassionate (Hayes, 2005). Participants were able to guess based on images of faces alone whether certain politicians were Democrat or Republican, stating that they rated images perceived as more powerful as Republican, and images perceived as warmer as Democratic (Rule & Ambady, 2010). Thus, it appears that liberal individuals tend to endorse emotional values and are stereotyped as emphasizing emotion more than their conservative counterparts.

A previous study investigated posing bias in relation to politics (Thomas, Loetscher, Clode, Nicholls, & Sirigu, 2012). In this study, images of politicians were collected and analysed for posing bias. These researchers found that conservative politicians were more likely to be portrayed with a leftward posing bias than their liberal counterparts. This was hypothesized to be due to conservatives more effectively using emotion in their campaigns, and due to the previous finding that conservative individuals have heightened emotional sensitivity (Thomas et al., 2012). It is important to note that our study differs from this one, as this previous study examined the way individuals posed, whereas we seek to investigate the perceptions of these poses. Thus, we expect our results to differ from this previous study, as ours concerns the perception of images of everyday individuals, rather than the production of images of politicians. It is also important to note that, to our knowledge, this is the only study done examining the relationship between pose direction and politics, and as it is an observational study which relies on samples of images that may suffer from selection biases, further research is needed to clarify this area.

The present study seeks to investigate how previously identified perceived differences in emotionality based on pose direction (Dunstan & Lindell, 2012; Lindell & Savill, 2010; Nicholls, Wolfgang, et al., 2002; ten Cate, 2002) affect ratings of perceived political orientation. We sought to explore this by presenting images of leftward and rightward posing individuals and asking participants to rate the perceived political orientation of that individual. Due to findings that leftward poses are rated as more emotional (Nicholls, Wolfgang, et al., 2002), and that liberals tend to be associated with greater emotionality (Farwell & Weiner, 2000; Hayes, 2005; Rule & Ambady, 2010; Winter, 2010), we hypothesize that individuals exhibiting a leftward posing bias will be rated as significantly more liberal than individuals posing with a rightward bias.

Methods

Participants

Eighty-two participants (40 male, 40 female, 2 gender-fluid) were recruited via an online advertisement at a Western Canadian university. Their mean age was 27.04 years (SD = 8.863, Range = 43).

Stimuli

Images of individuals posing with a leftward and rightward bias were collected from the FEI Face Database (available from http://fei.edu.br/~cet/facedatabase.html; Department of Electrical Engineering, Centro Universitario da FEI, 2006), a database available for research purposes that contains colour images of individuals with a neutral expression taken against a white backdrop with various face rotations. From this database, we collected images of 24 individuals (13 male, 11 female) exhibiting a leftward posing bias and rightward posing bias, for a total of 48 images. Each image was also mirror-reversed, for a total of 96 images used to create the online survey. Faces were of everyday individuals and included individuals of various ages and ethnicities, included the hair of the individuals, they included men with and without beards, and women with and without make-up. We selected these images because we wanted to balance naturalistic images (those that looked like everyday individuals) with control over pose direction (in this database, degrees of pose direction is specified with each image). Stimuli presented to participants were 300 × 225 pixels. Because participants completed the survey as an online questionnaire, viewing distance, viewing time and screen size were not controlled.

Measures

Political orientation

Participants rated each image for the perceived political orientation of the individual in the image. This dimension was rated on a 6-point Likert scale that ranged from “Very conservative” to “Very liberal”. Scores ranged from 1 to 6, where a higher score meant more liberal, and a lower score meant more conservative. Options were presented vertically in a radio button format to participants, with half of questions starting with “Very conservative” as the first option ranging to “Very liberal” as the last option, and the other half starting with “Very liberal” as the first option ranging to “Very conservative” as the last option. Each participant received an average score of their ratings of political orientation for men posing to the left, men posing to the right, women posing to the left, and women posing to the right. These averages ranged from 1 to 6.

Procedure

Participants completed an online survey. The survey contained four blocks of 24 photos of individuals posing to the right or left, of which each participant was randomly assigned to complete two blocks. Each block contained six original right pose images, six original left pose images, six mirror-reversed right pose images, and six mirror-reversed left pose images. Each block contained all 24 faces, and thus each participant saw the same model twice throughout the survey. Participants rated each photo for political orientation, and were allowed to view each image for as long as they needed to make a decision. After rating 48 photographs, participants completed demographic questions.

Results

In our analysis, we had two aims: (1) to examine whether the mean ratings of political orientation differed between posing bias and sex of the model, and (2) to determine whether the higher rating of political liberalism was due to pose direction, and not other pictorial cues or perceived differences in emotionality of the cheek that was being shown (i.e., pictures showing more of the left cheek might be seen as more emotional regardless of whether it is the original image or mirror-reversed). To examine this, a 2 × 2 × 2 repeated-measures ANOVA with within-subjects factors sex (2 levels: male, female), pose direction (2 levels: right, left) and reversal (2 levels: original, reversed) was conducted. There was a main effect of pose direction, such that individuals posing to the left (M = 3.71) were rated as significantly more liberal than individuals posing to the right (M = 3.64), F(1, 81) = 4.063, p = .047,  = .048. The main effect of sex was also significant, F(1, 81) = 40.50, p < .001,  = .33, such that females (M = 3.87) were rated as more politically liberal than males (M = 3.48) (see Figure 1

Figure 1. Mean perceived political orientation (±95% confidence intervals) by sex and political orientation. Higher scores indicate higher rated liberalism.

). However, the effect of reversal was not significant, F(1, 81) = .008, p = .931,  = .008. These results suggest that it is the face side that people are perceiving, and not the side of the face that the poser originally posed with, that determines rating of political orientation. There were no significant interaction effects. Distributions of our data by category are presented in Figure 2.

Figure 2. Distribution of the proportion of each response chosen split up by category (males posing to the left, males posing to the right, females posing to the left, females posing to the right).

Discussion

We sought to determine whether posing bias influenced perceptions of political orientation, and found that individuals with a leftward posing bias were rated as significantly more liberal than individuals posing with a rightward bias. This corroborates with previous researchers who have found that the direction of pose is related to perceived differences in emotionality (Dunstan & Lindell, 2012; Nicholls, Wolfgang, et al., 2002), and that these differences can effect perceptions of other aspects of the individual, such as how scientific the individual is (ten Cate, 2002) or what their major is in university (Lindell & Savill, 2010). These results also support the body of research that suggests that liberals are stereotyped as being more emotional than conservatives (Farwell & Weiner, 2000; Hayes, 2005; Rule & Ambady, 2010; Winter, 2010), as we found that the more emotional left cheek was associated with greater perceived liberalism. It is possible that participants in this study rated individuals posing with a leftward posing bias as being more liberal due to the association of liberalism with the “left” wing of politics, and those with a rightward posing bias as being more conservative due to the association of conservatism with the “right” wing of politics, suggesting a potential demand characteristic. It is possible that demand characteristics contributed to this effect, although even that effect would be reflective of the bias we were seeking to detect. Additionally, based on previous research that shows differences in perceptions of individuals based on pose direction where left and right metaphors are not involved (e.g., major in university, how scientific the individual is) (Lindell & Savill, 2010; ten Cate, 2002), there is reason to suggest that our results are similarly attributable to the posing bias, and not due to demand characteristics.

These results are in apparent contradiction to a previous study on posing bias and political orientation, which found that conservative political candidates were more likely to be portrayed with a more emotional leftward posing bias than their liberal counterparts (Thomas et al., 2012), until one considers several important differences between these two studies. Firstly, this previous study examined the way individuals posed, whereas we investigated perceptions of political stance based on pose direction. It is possible that politicians associated with liberalism or conservatism wish to use their political images to counteract the stereotypes of their parties, with conservatives choosing the more emotional left to counteract their stereotype as strong and authoritative yet emotionless leaders, and liberal politicians choosing the left less often to counteract their stereotype as emotional and empathetic yet weak leaders (Winter, 2010). Secondly, the previous study looked at political leaders, as opposed to the partisanship of everyday individuals. Future research should look at perceptions of political party membership based on posing bias using images of politicians to determine whether differences in perceptions exist between leaders and partisans. It could be that the perceptions of politicians simply do not align with perceptions of individuals who support either conservative or liberal politics.

Notably, our effect size for pose direction was quite small ( = .048). While our results suggest that the effect is small for a single trial and exposure of a single individual, the practical significance of our results should not be evaluated by looking at the individual only. While our effect is small for the individual, the contexts in which our study would be useful, such as political campaigns, involve large numbers of individuals. In this setting, there are many exposures, and thus it is the cumulative effect of these exposures that matter. In this context, even a small effect size can have a large impact (Fritz, Morris, & Richler, 2012). A related concern is that our findings are a statistical artefact, and that we did not have sufficient power to detect an effect. While our effect is small, there was a very small amount of variance in our study (see Figure 1), and thus it is the small amount of variance that accounts for our ability to detect a small effect.

Our results also showed that females were rated as significantly more liberal than males. This reflects the tendency of women to be more liberal than men in their voting behaviour and attitudes on political issues (e.g., see Chaney, Alvarez, & Nagler, 1998; Gidengil, Hennigar, Blais, & Nevitte, 2005; Kaufmann, 2002; McCue & Gopoian, 2000; Schlesinger & Heldman, 2001). It is also important to note that traits used to describe liberal individuals: compassionate, devoted to others, emotional, and kind, tend to be associated with femininity, and traits used to describe conservative individuals: active, independent, and decisive, tend to be associated with masculinity (Winter, 2010). This stereotyped association of liberals with traditionally feminine traits and conservatives with traditionally masculine traits could also have influenced participant ratings of individuals.

One major caveat to the current findings is their generalizability. The participants in the current survey were all affiliated with a Canadian university, suggesting exposure to North American politics, which makes use of the left-right political continuum. It is important to note that not all cultures have similarly organized political structures (Jou, 2010), and that often people use more than the left-right continuum when discussing political orientation (Duckitt, 2001; Krauss, 2006; Saucier, 2000). In fact, whether individuals perceive political orientation as a relevant dimension of a person is highly variable depending on the context (Skitka, Mullen, Griffin, Hutchinson, & Chamberlin, 2002). This means that there are likely certain populations in which there is no relationship between posing bias and perceived political orientation, as the political group with which a person identifies may not be perceived as definitive of their character. It is also possible that stereotypes of liberals and conservatives differ by culture, as the bulk of the research cited here focused on populations in North America. Future research should investigate effects of posing bias on perceived political orientation cross-culturally to determine whether differences exist, as well as investigate perceptions of opinions on particular political issues associated with each party, rather than judging partisanship as a whole.

For the purposes of this study, we assumed liberal and conservative political orientation were opposites along a single continuum. However, political orientation is often thought of as being more multidimensional than a range from conservative to liberal (see, e.g., Nolan, 1971). Future research should investigate whether individuals are, in fact, perceived on a single dimension from liberal to conservative, or whether individuals can be perceived as varying amounts of both liberal and conservative (i.e., utilize two separate scales to measure both perceived conservative orientation and perceived liberal orientation).

We found a difference in perceived political orientation based on posing bias, where individuals posing with the more emotional leftward posing bias were rated as more liberal than individuals posing with a rightward posing bias. This difference is likely due to the fact that liberal individuals are consistently stereotyped as more emotional than conservative individuals (Farwell & Weiner, 2000; Hayes, 2005; Rule & Ambady, 2010; Winter, 2010). These results support differences in perceived emotionality based on pose direction (Dunstan & Lindell, 2012; Nicholls, Wolfgang, et al., 2002) and support that a leftward posing bias is associated with characteristics of greater emotionality, such as a liberal political orientation.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This research was funded by the Natural Sciences and Engineering Research Council [grant no RGPIN-2015-04620].