Science & Society

Social attributions from faces bias human choices Christopher Y. Olivola1, Friederike Funk2, and Alexander Todorov2 1 2

Tepper School of Business, Carnegie Mellon University, Pittsburgh, PA, USA Department of Psychology, Princeton University, Princeton, NJ, USA

Our success and well-being, as individuals and societies, depend on our ability to make wise social decisions about important interpersonal matters, such as the leaders we select and the individuals we choose to trust. Nevertheless, our impressions of people are shaped by their facial appearances and, consequently, so too are these social decisions. This article summarizes research linking facial morphological traits to important social outcomes and discusses various factors that moderate this relationship. Throughout history and across cultures, individuals and societies have maintained the belief that the human face can reveal various aspects of a person’s true nature and intentions [1]. So prevalent were these beliefs that some scholars tried to develop a science of ‘physiognomy’ – a system for identifying personality types, or even criminals, from facial characteristics [1]. Although these attempts were eventually debunked and abandoned within the scientific community [1], the lay belief that faces are ‘windows to the soul’ has persisted. People continue to draw inferences about the characteristics of others from their facial appearances and these social attributions, in turn, can have important consequences. The outcomes of our most important social decisions (e.g., which political leaders to elect, which person to marry, etc.) depend on our ability to draw accurate inferences about other people’s tendencies, motivations, and qualifications. Unfortunately, these judgments are often influenced by superficial and weakly diagnostic cues. In particular, our impressions of people are heavily shaped by their facial appearances. The face is a rich source of information about a target person’s age, gender, ethnicity, and emotional state, yet face-based social inferences also fuel judgments about personality, behavioral intentions, and cognitive abilities [1]. As a result, facial traits can bias human choices. The potential for faces to influence social decisions is illustrated by their ability to predict a wide range of important outcomes. Face-based social attributions predict real-world outcomes A large body of research shows that facial appearances predict significant social outcomes in domains as diverse as politics [2–5], law [6–10], business [3,11], and the military [3,12]. Many of these studies find that individuals who Corresponding author: Olivola, C.Y. ([email protected]). Keywords: social judgment; facial appearance; person perception; facial stereotyping; human decision-making. 1364-6613/ ß 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.tics.2014.09.007

566

Trends in Cognitive Sciences, November 2014, Vol. 18, No. 11

possess particular facial characteristics are more likely to experience desirable outcomes (e.g., winning an election) or avoid undesirable outcomes (e.g., being convicted of a crime) than are their peers who lack these facial attributes. There is strong agreement across individuals when it comes to judging which faces look competent, trustworthy, and so on, and considerable progress has been made in identifying the configurations of facial features that lead to these inferences (Box 1). The literature on face-based inferences and their consequences is too vast to fully cover here, but two broad categories of decisions are worth discussing, given their importance to individual and societal wellbeing. The first category of biased decisions involves leadership selection and rank attainment. Numerous studies have shown that we form impressions of aspiring leaders from their faces, which in turn predict their success in reaching prestigious leadership positions. In the domain of politics, candidates’ chances of electoral success are related to how competent, dominant, sociable, threatening, conservative, and even ‘politician-like’ their faces make them look [2–5] (Figure 1). Within the domain of business, CEOs whose faces are perceived to look more competent are more likely to be hired by large, successful companies [11], even though they perform no better than their less competentlooking peers [1]. So strong, it seems, is this facial bias, that the CEOs of Fortune 500 companies have more competentlooking faces than other types of leaders [3]. Finally, within the military domain, having a face that is perceived to be dominant-looking predicts rank attainment [3,12]. The second category of biased decisions concerns judgments of trust, guilt, and criminality. Here the evidence shows that people are more likely to trust, and less likely to convict, individuals whose faces are generally perceived to look trustworthy or innocent. Financially, having a face that is perceived to be trustworthy-looking strengthens a person’s ability to attract investments and procure loans. One recent study found that potential borrowers on a peerto-peer lending site were more likely to have their loans funded if they had a trustworthy-looking appearance [13]. This finding is further corroborated by experimental laboratory studies showing that participants in strategic economic games are less willing to trust and invest money in partners who have untrustworthy-looking faces, even when relevant information about their past behaviors is available [14,15]. The implications for the legal system are no less serious because facial appearances predict judgments of criminal guilt. Individuals who have stereotypically criminal-looking faces are more likely to be selected from a police lineup, and thus to face trial [8]. Once on trial, defendants who have untrustworthy-looking faces [9] or faces that fit the

Science & Society

Trends in Cognitive Sciences November 2014, Vol. 18, No. 11

Box 1. Building models of social attributions from faces Psychologists have known that there is consensus in social attributions from faces since the early 20th century [1]. Relying on this consensus has been the most common approach to measuring the impressions that faces convey. In a typical study, participants are asked to evaluate faces on the dimension(s) of interest (e.g., competence) and their aggregated judgments are then used to predict an outcome (e.g., electoral success). Alternatively, researchers can preselect faces rated high (vs low) on the dimension(s) of interest and use these faces to study the effects of appearances on decisions in experimental studies. Recently, new methods have been developed to build computational models of this consensus and visualize the configurations of facial features that lead to specific social attributions [1]. These datadriven methods start with a mathematical representation of facial morphology in which each face is completely characterized by a set of vector coordinates. Using people’s evaluations (e.g., of competence, trustworthiness, etc.) of randomly generated faces from this

representation, it is possible to build a facial-morphological model of the social dimension(s) being evaluated. These models visualize the changes in the face that contribute to specific social attributions. Figure I shows a computerized face manipulated to look more (or less) competent, dominant, extroverted, and trustworthy. The changes in the face are holistic – that is, not limited to single features – and capture the systematic variance in social attributions from faces. The trustworthiness model, for example (Figure ID), shows that faces perceived to be trustworthy-looking tend to be feminine and resemble the expression of positive emotions, whereas untrustworthy-looking faces tend to be masculine and resemble the expression of negative emotions. These models can be applied to novel faces, which in turn can be used to experimentally study the biasing effects of facial morphology on social choices. Several research groups have used model-derived faces to demonstrate the causal influence of facial morphology on human decision-making [1,15].

(A)

(B)

(C)

(D)

TRENDS in Cognitive Sciences

Figure I. Faces generated by data-driven computational models of evaluations of (A) competence, (B) dominance, (C) extroversion, and (D) trustworthiness. The face in the middle column represents an average face in the statistical model. Faces in the right column are 3 standard deviations (SD) above the average face on the respective trait dimension; faces in the left column are 3 SD below the average face on that same dimension [1].

567

Science & Society

Candidate vote share

(A)

Trends in Cognitive Sciences November 2014, Vol. 18, No. 11

Low polical knowledge voters

(i) 1

1

0.9

0.9

0.8

0.8

6***

) = .3

r (128

0.7

0.7

0.6

0.6

0.5

0.5

0.4

0.4

0.3

0.3

0.2

0.2

0.1

0.1

0

0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Likelihood of being rated more competent-looking than rival

Democrat candidate vote share

(B)

(i)

0.1

0.2

(ii)

Republican parcipants

0.9

0.9

0.8

0.8

0.7

4**

7) = .2

r (16

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Democrat parcipants

0.7 0.6

0.5

0.5

0.4

0.4

0.3

0.3

0.2

0.2

0.1

0.1

0

r (167) = –.

13

0 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Likelihood of Democrat being (mis)idenfied as Republican

Winning vote share

0

1

(C)

r (120) = –.06

Likelihood of being rated more competent-looking than rival

1

0.6

High polical knowledge voters

(ii)

(i)

0

0.2

(ii)

Republican policians

0.8

0.8

0.7

0.7

r (35) = –.02

0.6

0.1

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Likelihood of Democrat being (mis)idenfied as Republican

Democrat policians

0.6

0.5

r (26

0.5

0.4

)=–

.45*

0.4 0.3

0.4

0.5

0.6

0.7

0.8

Likelihood candidate is correctly idenfied as a polician

0.3

0.4

0.5

0.6

0.7

0.8

Likelihood candidate is correctly idenfied as a polician TRENDS in Cognitive Sciences

Figure 1. Various face-based social judgments of political candidates can predict their popularity among voters, but the strength of this relationship often depends on one or more moderating factors. (A) Judgments of competence from candidates’ facial photos (specifically: the proportion of participants who judged that these candidates look more competent than their rivals) positively predict their vote shares in US Senate and Gubernatorial elections [5], but this relationship is moderated by voter political knowledge [2]: (i) the relationship is particularly strong for voters with limited political knowledge (collapsing across levels of television exposure). (ii) By contrast, the voting preferences of voters with substantial political knowledge are not predicted by candidate facial competence (collapsing across TV exposure). (B) How stereotypically Republican-looking candidates are (specifically: the proportion of participants who categorized them as Republicans solely from their photos) positively predicts their vote shares in US Senate and Gubernatorial elections, but this relationship is moderated by the political leaning of voters [4]: (i) political facial stereotyping strongly predicts the voting preferences of Republican voters. (ii) By contrast, it does not predict those of Democrats. (C) Looking stereotypically like a politician (specifically: the proportion of participants who correctly categorized candidates as political leaders, rather than other types of leaders, solely from their faces) is negatively related to the winning vote shares obtained by US Gubernatorial election winners, but only for some candidates [3]: (i) the winning vote shares of Republicans are unrelated to how stereotypically politician-like they look. (ii) By contrast, it negatively predicts the margin of victory among Democrats. (A) adapted from [2,5], (B) adapted from [4], and (C) designed using data drawn from [3]; *, p <0.05; **, p <0.01; ***, p <0.001.

568

Science & Society stereotype of the crime they are being tried for [7] are more likely to be found guilty. Babyfacedness (i.e., childlike facial features) also predicts judgments of guilt and punishment severity [6,10]. More than just correlations: the causal evidence Faces are highly complex stimuli, and are thus difficult to experimentally manipulate. Consequently, most of the research linking social attributions from faces to decision making has been correlational in nature. Fortunately, many researchers have risen to the challenge and carried out controlled studies of face-based social inferences. These researchers have generally taken one of two approaches: some have used computer software to systematically manipulate the facial appearance of target individuals [1,15] (Box 1), while others have capitalized on natural variations in appearance by having normal faces pre-rated on several dimensions (e.g., competence, trustworthiness, etc.) and then selecting those faces that were rated high or low on the dimensions of interest for further study [6,7,9,14]. By systematically altering or selecting the faces that participants are presented with, experimenters have been able to examine how variations in facial appearance bias human decisions. These studies have provided causal evidence that facial appearances influence voting preferences [1], economic exchanges [1,14,15], and legal judgments [1,6,7,9]. Moderators and boundary conditions Critically, the predictive power and impact of face-based social judgments are not invariant, but instead depend on several factors. One important factor is the nature of the social decision that a human judge is faced with making. Among other things, this determines the traits considered most relevant to the decision at hand, and thus most likely to be evaluated from faces. For example, people tend to associate political leadership with competence, and military leadership with masculinity, maturity, and low warmth [3]. Not surprisingly then, having a face that is generally perceived to look competent is one of the strongest appearance-based predictors of success and status in the political domain [5], whereas in the military having a face that is perceived to be dominant-looking is a better predictor [12]. The relationship between appearances and social choices also depends on the characteristics of the decision-makers (Figure 1). Several studies suggest, for example, that voters differ in how strongly they are influenced by candidates’ facial appearances. One found that Republican voters are more likely to vote for candidates who have stereotypically Republican-looking faces, whereas Democrats do not show a preference for Democrat-looking (or Republican-looking) candidates [4] (Figure 1B). Another study showed that voters who have limited political knowledge are more likely to vote for candidates with competent-looking faces (all else being equal), whereas voters with high political knowledge show no such tendency [2] (Figure 1A). This last study illustrates another important factor, which is access to information. Providing decision-makers with relevant information can sometimes mitigate the biasing impact

Trends in Cognitive Sciences November 2014, Vol. 18, No. 11

of facial appearances. For example, people rely less on facial trustworthiness to make investment decisions when they have access to information about relevant past behaviors [14,15]. Unfortunately, even when they possess clearly relevant information, people can still be influenced by facial appearances [4,7,14,15]. What to do about face-ism The fact that social decisions are influenced by facial morphology would be less troubling if it were a strong and reliable indicator of people’s underlying traits. Unfortunately, careful consideration of the evidence suggests that it is not [1]. The accuracy of face-based inferences often drops once the target person’s gender, ethnicity, and age are controlled for [1,4]. Moreover, the reliance on facial information leads observers to neglect other, more valid, cues, which ultimately harms their judgmental accuracy [1]. Studies have also shown that human observers have surprisingly weak meta-accuracy: they are unable to properly evaluate how well (or not) they can judge social characteristics from faces [1,3]. Finally, recent evidence shows that people form very different social attributions from distinct photos of the same individual [1], which clearly contradicts the notion that stable morphological facial features drive accurate and reliable social inferences. Therefore, researchers and policy makers should strive to reduce the biasing impact of appearances on human judgments and choices. This is a challenging task because people are naturally inclined to draw inferences from faces [1,5] to an extent that they may find it difficult to inhibit these tendencies. On the positive side, the evidence suggests that people sometimes rely on facial appearances less when they are armed with more relevant and valid types of information [2,15]. Thus, in some contexts, educating people might be sufficient to reduce facial stereotyping. In other contexts, however, more research will be necessary to identify the best ways to mitigate the biasing influence of facial appearance. For instance, it still remains to be determined how justice can be truly blind – that is, how judges and juries can disentangle case-relevant facial information (e.g., expressions of remorse) from information that should be irrelevant to a case (e.g., facial morphological features perceived as criminal-looking). References 1 Todorov, A. et al. (2015) Social attributions from faces: determinants, consequences, accuracy, and functional significance. Annu. Rev. Psychol. 66, 2015 2 Lenz, G.S. and Lawson, C. (2011) Looking the part: television leads less informed citizens to vote based on candidates’ appearance. Am. J. Polit. Sci. 55, 574–589 3 Olivola, C.Y. et al. (2014) The many (distinctive) faces of leadership: inferring leadership domain from facial appearance. Leadersh. Q. 25, 817–834 4 Olivola, C.Y. et al. (2012) Republicans prefer Republican-looking leaders: political facial stereotypes predict candidate electoral success among right-leaning voters. Soc. Psychol. Pers. Sci. 3, 605–613 5 Olivola, C.Y. and Todorov, A. (2010) Elected in 100 milliseconds: appearance-based trait inferences and voting. J. Nonverbal Behav. 34, 83–110 6 Berry, D.S. and Zebrowitz-McArthur, L. (1988) What’s in a face? Facial maturity and the attribution of legal responsibility. Pers. Soc. Psychol. Bull. 14, 23–33 569

Science & Society 7 Dumas, R. and Teste´, B. (2006) The influence of criminal facial stereotypes on juridic judgments. Swiss J. Psychol. 65, 237–244 8 Flowe, H.D. and Humphries, J.E. (2011) An examination of criminal face bias in a random sample of police lineups. Appl. Cogn. Psychol. 25, 265–273 9 Porter, S. et al. (2010) Dangerous decisions: the impact of first impressions of trustworthiness on the evaluation of legal evidence and defendant culpability. Psychol. Crime Law 16, 477–491 10 Zebrowitz, L.A. and McDonald, S.M. (1991) The impact of litigants’ baby-facedness and attractiveness on adjudications in small claims courts. Law Hum. Behav. 15, 603–623

570

Trends in Cognitive Sciences November 2014, Vol. 18, No. 11 11 Rule, N.O. and Ambady, N. (2008) The face of success: inferences from chief executive officers’ appearance predict company profits. Psychol. Sci. 19, 109–111 12 Mueller, U. and Mazur, A. (1996) Facial dominance of West Point cadets as a predictor of later military rank. Soc. Forces 74, 823–850 13 Duarte, J. et al. (2012) Trust and credit: the role of appearance in peerto-peer lending. Rev. Financ. Stud. 25, 2455–2484 14 Chang, L.J. et al. (2010) Seeing is believing: trustworthiness as a dynamic belief. Cogn. Psychol. 61, 87–105 15 Rezlescu, C. et al. (2012) Unfakeable facial configurations affect strategic choices in trust games with or without information about past behavior. PLoS ONE 7, e34293

Social attributions from faces bias human choices

ders we select and the individuals we choose to trust. Nevertheless, our impressions of people are shaped by their facial appearances and, consequently, so too ...

843KB Sizes 0 Downloads 204 Views

Recommend Documents

Social attributions from faces bias human choices
1 Tepper School of Business, Carnegie Mellon University, Pittsburgh, PA, USA. 2 Department of ... and emotional state, yet face-based social inferences also.

Religious Attributions, Secular Attributions, and Religious Priming.pdf
God May Save Your Life, but You Have to Find Your Own ... tions, Secular Attributions, and Religious Priming.pdf. God May Save Your Life, but You Have to Find ...

A Generic API for Retrieving Human-oriented Information from Social ...
A Generic API for Retrieving Human-oriented Information from Social Network Services. Teruaki Yokoyama, Shigeru Kashihara, Takeshi Okuda, Youki Kadobayashi, and Suguru Yamaguchi. Graduate School of Information Science, Nara Institute of Science and T

choices, choices
A look at the pros and cons of various types of Web advertising. By JENNIFER REWICK. For many marketers, the big question isn't whether to adver- tise on the ...

éBIAS/
Nov 13, 1995 - output signal of the photo-detector increases with an increas. U'S' P ...... digital measurement signal produced at the output 110 of the counter.

Bias Neglect
community is that there should be less aggression between ants that share a nest ... condition. Blind studies were much more likely to report aggression between.

A Novel Approach for Recognition of Human Faces
Email: * [email protected] , ** [email protected]. ABSTRACT: This paper attempts to describe ... template based methods. Here we are going to recognize human ...

Protein Folding and Ligand-Enzyme Binding from Bias ...
In Section II, the convergence criteria, and the methods for analyzing data to ..... [19] a formalism was introduced which allows to map the history-dependent.

attributions and outcomes of the
how to get things straightened out"), the contact person (“that young woman ... The customer calls the. Credit Card .... The subjects were from an advanced marketing .... (either a file or box depending on the particular center) that was checked .

Confusion about concessive knowledge attributions
Feb 5, 2008 - discussed recently. It's agreed by all that CKAs are infelicitous, at least typically. ... of all sorts of philosophical theses; to name just a few: contextualism, invariantism, fallibi- ...... Social Epistemology (forthcoming a). Schaf

attributions and outcomes of the
behavioral intentions are more favorable in stable service recoveries, 2) an employee based service recovery ... It is widely recognized that no service system is perfect. ...... Flight routes to some destinations are virtual monopolies, ... direct m

Design Choices in Privatized Social-Security Systems
In the 2000 U.S. Presidential campaign,. George W. Bush ... launching a system that is very similar to Pres- .... equipped to choose what is in their best interest,.

Bias Neglect
Experimenter bias occurs when scientists' hypotheses influence their results, even if involuntarily. Meta-analyses (e.g. ... This is true even when participants read descriptions of studies that have been shown to ... sometimes influence their result

éBIAS/
Nov 13, 1995 - source to cancel gain changes produced by changes in ambient ..... The analog output signal of the peak averager and memory circuit 90 is.

-1cm Advertising Spending and Media Bias: Evidence from News ...
and participants in the 2016 Media Economics Workshop, the 10th Toulouse Internet ... this case illustrates the importance of media scrutiny and the potential social ... focusing on the top 100 recalls in terms of the number of potentially affected .

Social competition and firms' location choices
fi and social taxes τi. Moreover, the equilibrium wage resulting from the Nash-bargaining process (w q i ) depends on the union's bargaining power βi, unemployment benefits bi, the firing cost fi and the degree of centralization of the wage-bargain

Confusion about concessive knowledge attributions
Feb 5, 2008 - expressions. Lewis (1996) motivated his own version of contextualism by its ability ... For they want many of our knowledge attributions to ... in C, and that explicitly referring to a possibility in an utterance will make that pos-.

Fallibilism and concessive knowledge attributions
ANALySiS 65.2, April 2005, pp. 126–31. © Jason Stanley .... of epistemic possibility, allows for an explanation of the data in (1). The sentences in (1) are ...

Causal Attributions, Perceived Control, and ...
Haskayne School of Business, University of Calgary, 2500 University Drive, NW, Calgary,. Alberta ..... Working too hard .81 .13 .59 .27. Depression .60 .05 .74. А.13. Not doing enough exercise .49 .15 .64 .06. Working in an environment with no fresh

Protein Folding and Ligand-Enzyme Binding from Bias ...
Keywords: Enhanced sampling, Free energy calculations, Protein folding, Ligand-enzyme binding, ..... history dependent potential according to Eq. 1, but then sets. 0. 50. 100. 150 ...... LysM domain using a coarse-grained model. J. Phys.

Multivariate choices and identification of social ... - Wiley Online Library
Jun 1, 2017 - This paper considers the identification of social interaction effects in the context of .... We keep track of social connections in the network.