First impressions formed from faces
This column explores the tendency to form judgments about a person from our first impressions based on their facial features. It aims to provide a definition of these first impressions from a face and seeks to show how these can influence our judgment.
In psychological research, ‘first impressions from faces’ refers to characteristics (e.g., personality traits) that we attribute to someone based on looking at their face. When we see a new face, we all form first impressions and tend to assume they are true, albeit unconsciously.
You decide to participate in a study in which you are asked to guess a person's occupation from a photo. The research team presents the following portrait (Figure 1).
From the list, what do you think this person's job is?
Figure 1 Face presented during experiment, from Todorov (2017).
Based on this person’s clothing, you may have excluded the responses "farmer", "luthier", "carpenter", even "salesman" and you deduced that this gentleman probably belongs to a qualified professional category with a university degree. You probably also relied on looking at that person's face to formulate your answer [1, 2].
Beyond the question posed, it is likely that when you saw this face, you formed a particular impression of this man: you may have found that he looked rigorous, severe, or competent, for example [2, 3].
Maybe you recognized this face? This is President Warren Harding, 29th President of the United States. At that time (1920s), physiognomy - that is, the study of faces to determine a person's character - was in vogue and practiced by physiognomists. Physiognomists of the time described Harding as a man of spirit, with a scientific intellect. It was predicted then that he would be the best president the United States had ever had. Yet historians today consider him one of the worst presidents of the United States: debauchery, corruption, scandals... .
This example illustrates that when we perceive a face, we tend to generate impressions from it: we attribute traits and characteristics (e.g., personality traits) to a person from their facial features. Additionally, the example shows that we use these impressions to make judgments (e.g., "He looks competent, severe, so he must have a job that involves heavy responsibilities").
Heuristics and first impressions
The example of President Harding illustrates how we use our impressions from a face to answer complex questions. To answer these questions, we use a heuristic . A heuristic can be defined as a "shortcut" used in order to answer a complex question. In the context of President Harding's photo, we do not have enough information available to formulate a definite answer as to his profession. So, to answer the question, we use the elements that are at our disposal: his clothing and the impression that his face generates. This example shows us how heuristics allow us to form judgments based only on information that is immediately accessible.
Indeed, information is susceptible to be used in heuristics when it is directly accessible in our environment or in our memory. Thus, when we use heuristics, it is likely that we are using this easily accessible information to form our judgment .
Certain types of information are considered to be systematically easily accessible, that is, they are such that it is not necessary to make an effort to access them. This includes the information that we perceive with our senses, as well as the information that we have stored in memory .
First impressions are a particularly accessible category of information . Indeed, for information processed by our brain to reach our consciousness, we generally consider 120 milliseconds to be necessary . However, when we study the time it takes to form a first impression from a face, only 34 milliseconds are needed, suggesting that first impressions are not a conscious phenomenon . In addition, we are not able to prevent the production of these first impressions. This suggests that impression formation is a process that occurs quickly and automatically. Thus, first impressions are particularly accessible information that we can call on when we are using heuristics to form a judgment.
When we are exposed to a face, we quickly form a first impression of it. We would produce these first impressions automatically, without being aware of it, and we would not be able to inhibit them. In addition, they would be particularly accessible, that is to say that we can have access to them without having to provide a cognitive effort. They could then be used as a heuristic to answer complex questions.
What determines first impressions?
Do facial features really reveal a person's personality? This belief has gone through history: from Aristotle to 18th century physiognomists .
Faces, and especially the information they convey, attracted the attention of early research in psychology, both in the detection of emotions and the inference of personality traits. Today, research has shifted away from the idea that the face reflects a person's internal characteristics, other than their emotions. Taking for granted that faces do not reflect personality, research focuses instead on the mechanisms that lead to formulating our judgments from the facial features. We are here talking about impression formation [5, 8].
In the following sections, we discuss the social and perceptual origins of first impressions from faces. First impressions from faces are both a perceptual phenomenon (that is to say, generated through our senses) and a social phenomenon, for their meaning and influence are, in part, learned and shared within social groups.
Some theories of perception suggest that we perceive in order to act . Thus, our perception would not simply transcribe the outside world but would allow us to identify certain information used to produce an action. Some researchers investigate the perception of the social world. These researchers are specifically interested in information about a person's appearance, the way they speak, and so on, which communicate their intentions or their emotions, but which also qualify the relationship we have with them (e.g., friendship, family, professional relationship, etc.). In this perspective, faces would communicate information that would guide action (e.g., we perceive a frightened face and therefore are on guard). Although we are discussing here only the perception of facial features, it is important to note that faces are not the only possible source of information which guide our actions during social interaction.
When we perceive a face, we can distinguish different types of information that tell us about the person and the situation (for example, age group, membership of a population group, or emotional expression) . This ability to detect and interpret information from a face may be innate but is also acquired over the course of life. For example, children are particularly sensitive to the faces of adults, while adolescents and adults are more sensitive to the faces of their peers . In particular, from adolescence onwards, information related to the detection of peers is prioritized, while in children, it is the detection of caregivers (for example, parental figures) that takes priority. Indeed, since children are still largely dependent on their caregivers, this prioritization of information would be adaptive: in the event of a problem, the child will probably need an adult. Being able to quickly identify an adult is therefore a priority. This sensitivity to adult faces can save the child time, which is crucial in dangerous situations. However, from adolescence onwards, children become empowered and tend to rely more on their peer group than on their caregivers. The development of a greater sensitivity to the faces of peers is associated with a greater inclination towards finding and identifying people you can trust, with whom to form loyal friendships, as well as future romantic partners.
Thus, during an interaction, the face can provide information that helps us understand the context of our interlocutor and facilitates communication. In summary, the information we extract and interpret from the face may depend on our age (for example, baby versus adolescent) but also on our social goals (for example, need for help) .
This sensitivity to certain facial cues can lead to misperceptions when the information we are exposed to is incomplete or impoverished. Indeed, we may be faced with situations in which the information available in the environment is partial or incomplete and where we nevertheless want to make a certain judgment. For example, you perceive that your friend seems sad. However, you don't have any information other than his facial expression to justify your impression. You are then in a situation in which your information is incomplete for the purposes of formulating your judgment. We then speak of impoverished information .
The appearance and maintenance of impressions from faces in contexts where information is impoverished can be adaptively explained: if you see someone running towards you who seems frightened, you will tend to be on your guard. If nothing happens, then your reaction will be of no consequence. However, if something does happen, you will need less time to react than if you had not considered your impression. From this perspective, it is advantageous (i.e., adaptive) to produce quick judgments, even if the information is incomplete, since the consequences of not making the judgment about a dangerous situation could be disastrous, even fatal.
However, situations in which information is depleted can lead to biased perceptions. Information that is usually adequately perceived from faces can lead to a phenomenon of overgeneralization. For example: the modulation of our behavior towards a baby is adaptive, that is to say that when we are in front of a baby, we tend to adopt a protective position, to inhibit our aggressiveness. This modulation is adaptive because it promotes the proper development of the child (and more generally his survival). However, in order to be able to modulate our behavior, we must be able to recognize the information that indicates that we are dealing with a baby. Among other things, the face allows this recognition . The perception of certain key information about the faces of babies, such as large eyes, the proportions of the face or the shape of the head would then trigger protective behaviors and inhibition of aggression. The reaction produced by babies’ faces is said to be so strong that it could also occur when perceiving some of this information on adult faces. This is the phenomenon of overgeneralization since the impression caused by features is generalized from infants to adults. Thus, an adult face that presents facial characteristics similar to those of a baby could induce the perception of a vulnerable or more submissive person, just as is the case when we perceive a baby's face.
These first explanatory elements allow us to understand the social origin of first impressions: the information we extract from faces guides our action during social interactions. More specifically, the interpretation of information deduced from faces allows us to attribute certain characteristics to others (e.g., intentions, emotions, but also the nature of the relationship maintained). The meaning of this information can be innate but also acquired during our lifetime and according to our social goals .
There is a certain consistency in our judgments , that is, it is thought that we all generally have the same first impressions from a face. This consistency suggests that beyond the social significance of the information we extract from faces, first impressions are also linked to perceptual phenomena. Thus, certain facial configurations would systematically cause certain first impressions.
Some researchers have tried to determine what these facial configurations are. One of the questions that interests them is: what are the facial characteristics that induce a perception of the face as being more dominant or more reliable? Researchers have identified factors that may influence dominance and reliability judgments produced by first impressions of a face. They found that by varying the characteristics of dominance and reliability they could generate a particular impression, as well as the intensity of the impression. For example, a square jaw has been shown to induce a stronger impression of dominance, so by varying the shape of the jaw (more or less square), we vary the impression of dominance .
The variations in impression that we are talking about are called dimensions. Thus, a dimension represents the variations of facial characteristics that determine an impression, to vary its intensity.
The figure below shows faces that vary on the reliability dimension. Who would you prefer to lend your car keys to?
Figure 2 - Reliability dimension (from less to more reliable). From Oosterhof and Todorov (2011)
Chances are, your choice is on faces on the right.
To build their models, the researchers used computer-generated faces that are emotionally neutral. Neutral faces are constructed from the facial features of people when they do not express emotion. It is important to note that just because you are not expressing emotion, it does not mean that you will be seen as not expressing any (this is the whole point of impressions from a face).
By generating a large number of neutral faces and asking participants to rate them according to the dimension of interest, we can determine the facial characteristics that cause judgments to vary. Once we know these characteristics, we can take a neutral face, and adjust its facial characteristics to vary the impression it generates. Figure 2 shows this last step.
Thus, the assessment of reliability would be based primarily on the perception of characteristics resembling emotional expressions. The angrier the face is perceived to be, the less reliable it appears (left face in the figure). Conversely, the more the face is perceived as happy, the more it is categorized as reliable (face on the right in the figure) . It would be difficult to make an exhaustive list of variations in facial features. However, it can be noted that the shape of the mouth is not the same for each of the faces: the more reliable the face is perceived to be, the more the line between the lips appears to have a "u" shape, and vice versa. This u-shaped line on the reliable face resembles that of a smiling person, so we perceive that face to be that of a happy person. This is the phenomenon of overgeneralization of emotional expressions that was mentioned above: we perceive facial features (for example, the line of the lips in the shape of a "u") that resemble emotional expressions (i.e., the line of the lips of a happy person) then we generalize them as if they were true (i.e., that person is happy).
Now let's take a look at the dimension of dominance. In the figure below: who would you hire to be your bodyguard?
Figure 3 - Dominance dimension (from less to more dominant). From Oosterhof and Todorov (2011)
Again, your choice will certainly be on the right face.
A male and mature face will be perceived as dominant, while a female face with childish features will be perceived as not very dominant (left face in the figure) . Dominance being associated with physical strength; the perception of dominance would indicate the capacity of people to harm us .
To date, seven other dimensions have been studied in addition to dominance and reliability: threat, attraction, fear, meanness, extroversion, competence, and agreeableness .
Accuracy of social attributes from faces
So far, we have studied the processes that lead us to formulate our impressions of faces. However, we have not looked at their accuracy: are our first impressions reliable?
While we are generally able to determine obvious information from data visible on faces (population group, age, etc.), impressions of features that are not objectively identifiable are far less accurate (e.g. personality traits) .
However, the debate is not completely settled. Some studies claim that we are able to determine political inclination, sexual orientation, criminality, and other internal traits from photographs of faces. These studies seem to suggest that faces would reveal characteristics of our internal being. However, these projects are criticized for their methodology which often has significant flaws . For example, some studies leave certain factors visible, such as the clothing of people in the photos they present. Consequently, the participants are exposed to multiple sources of information: the face and the clothing. Yet we know that factors like clothing would be enough to infer a person's class membership or political orientation, without even having to look at the face. Thus, if the judgments of the participants are accurate, it is not possible to know if this is caused by the faces or by the dress. For the moment, experts on the subject consider that there is no adequate evidence to show that impressions from faces are a reliable source of information for detecting characteristics such as personality traits (e.g., reliability, aggressiveness, agreeableness), skill level, political or sexual orientation, etc.
We might think that if our first impressions from faces were valid, then we should be able to put them to good use where relevant [5, 11]. For example, consider a scenario where we are asked to determine the political orientation of certain people from a photo. To do this, we are informed of the distribution of political orientations in the group of photos that we will see (90% Democrats vs. 10% Republicans). We should be able to classify at least some of the photos without difficulty: if we know that the group is 90% Democrats and we always answer ‘Democrat’ then we should have 90% correct answers. However, such results are very rare. On the contrary, participants in similar studies achieve much lower categorization rates than if they had used the fixed response strategy. This suggests that participants use their first impressions too heavily in formulating their judgment and that these mislead them. Objective information, such as whether someone belongs to certain demographic group (e.g., age group), is relatively accessible in our daily lives. It provides information that is more reliable than our first impressions. Thus, we would be better off neglecting our first impressions to focus on more reliable indicators .
In addition, our ability to judge the accuracy of our judgments - called meta-accuracy - seems limited. Indeed, if we ask participants to categorize faces according to occupations and then we ask them to judge their own performance, people fail to accurately predict for which photos they likely gave a correct answer . Additionally, participants who were more confident in their overall performance did not fare better than those who were less confident. This low level of meta-accuracy suggests that we are unable to detect the accuracy of our first impressions even when they lead us to a false judgment .
Ultimately, even if we succeed in obtaining evidence that certain facial characteristics are associated with certain personality traits, the evidence would not necessarily demonstrate the existence of a causal link between personality and facial morphology . Suppose that it has been shown that people perceived to be dominant more often exhibit violent behavior. This would not necessarily demonstrate a personality trait, since their violence could be explained by the behaviors adopted by others in their company. In other words, if we perceive a dominant face, we might tend to react more aggressively to the perceived threat, thus "forcing" the person to respond violently. In this way, the stereotype that associates the dominant face with violent behavior would be confirmed (this is the notion of self-fulfilling prophecy ) .
So, while the debate continues, it seems that we should not rely on our first impressions from facial features to infer internal characteristics of individuals. Nevertheless, it seems that these impressions can have an important influence in our judgments. In the next part, we detail the consequences of first impressions from faces in our daily life.
Our first impressions are particularly accessible during heuristic judgment, and we tend to rely on these in forming judgments. The formation of our first impressions has social and perceptual origins: when we are exposed to impoverished information, we overgeneralize our knowledge (for example, our knowledge of faces expressing emotions) to people with similar facial characteristics. In addition, certain perceptual determinants, and more precisely certain facial configurations, generate particular impressions (for example, dominance and reliability). And finally, the judgments we make based on our first impressions are unreliable, even biased.
Social consequences of judgments from faces
If our first impressions from faces are unreliable, then we should question their consequences and ask ourselves how much these influence our daily judgments.
In 2005, Todorov and his colleagues  tested the hypothesis that first impressions from faces affect social events. To do this, the researchers asked participants who did not know the candidates (two men) - to judge the competence of finalists in the US Senate elections of 2002 and 2004 from their faces. The results are astounding: in 72% of cases, the candidate judged to be the most competent was the current election winner - and that judgment was based solely on his face! Another surprising point is that the greater the difference in perception of competence between the candidates, the greater the difference between the votes. It is important to stress that this study does not address the competence of the candidates but rather the influence of the first impressions from faces on the choice of the voters. Thus, this study suggests that our first impressions can influence important decisions such as the choice of our political representatives. These results have been reproduced several times and reinforce these conclusions .
Beyond the election results, many studies have been conducted to understand the consequences of first impressions from faces. Here is a summary of some of their findings:
Military rank is correlated with the impression of dominance of the face .
People who appear more competent and dominant come more easily to the head of large companies and receive higher salaries than their colleagues, even if they do not perform better .
Participants will be less likely to trust people whose faces appear less reliable, even if their behaviors indicate that they are reliable .
From another perspective, with more important consequences, the reliability of the face of an accused influences the decision of the jurors responsible for judging his guilt. The effect is maintained even among participants endorsing the values of justice and equity .
Finally, the reliability of the patient's face significantly varies the inclination of healthcare professionals to take care of these patients .
On a more positive note, first impressions have less influence on the judgments of professionals with more experience than novices . This last point allows us to qualify the impact of first impressions in our daily life. Indeed, whether for health professionals or voters, it seems that expertise diminishes the influence of first impressions. If we take the example of elections, it seems that people with knowledge of politics are less likely to be biased by their first impressions from faces .
We all know the adage: "clothes don't make a man" or "you shouldn't judge a book by its cover". However, despite these warnings, it seems we cannot resist the temptation to attribute characteristics to others based on their appearance. The first impressions based on facial features are proof of this.
In this column we have seen how first impressions from faces can interfere in our judgments and have consequences in our lives. Specifically, we saw how the first impressions from faces are accessible when making a heuristic judgment: when we have to answer a complex question (for example, who should I vote for?), we rely on these impressions to formulate our judgments even if they are doubtful. Finally, their consequences can be significant, as we saw above with the example of elections.
It seems important to stress that these processes are automatic, unconscious, and shared by all. We cannot control them, but we can be aware of their existence. So how do we try to reduce the influence of first impressions from faces on our judgments? As explained above, it seems that expertise reduces bias. So, failing to be experts, we can make sure that we base our choices, such as our voting behavior, on objective evidence, such as political agendas.
 Zebrowitz, Leslie A. (1998). Reading Faces Window to the Soul. Westview Press.
 Todorov, Alexander B. (2017). Face Value the Irresistible Influence of First Impressions. Princeton University Press.
 Willis, Janine & Alexander B. Todorov (2006). First impressions. Psychological Science 17(7), 592–98.
 Kahneman, Daniel & Shane Frederick (2002). Representativeness revisited: Attribute substitution in intuitive judgment. In Gilovich Thomas & Dale Griffin (Eds.). Heuristics and Biases: the psychology of intuitive judgement, 49-81. Cambridge University Press.
 Todorov, Alexander, Christopher Y. Olivola, Ron Dotsch & Peter Mende-Siedlecki (2015). Social attributions from faces: Determinants, consequences, accuracy, and functional significance. Annual Review of Psychology 66(1), 519-545.
 Lamme, Victor A. (2010). How neuroscience will change our view on consciousness. Cognitive Neuroscience 1(3), 204–220.
 Todorov, Alexander, Manish Pakrashi & Nikolaas N. Oosterhof (2010). Evaluating faces on trustworthiness after minimal time exposure. Social Cognition 27(6), 813–33.
 Zebrowitz, Leslie A. (2011). Ecological and social approaches to face perception. In Rhodes Gillian, Andy Calder, Mark Johnson & James Haxby (Eds.). Oxford Handbook of Face perception, 1-20. Oxford Handbooks Online.
 Picci, Giorgia & Suzanne Scherf (2016). From caregivers to peers: Puberty shapes human face perception. Psychological Science 27(11), 1461-1473.
 Oosterhof, N. Nikolaas & Alexander Todorov (2008). The functional basis of face evaluation. Proceedings of The National Academy of Sciences 105(32), 11087-11092.
 Olivola, Christopher Y. & Alexander Todorov (2010). Fooled by first impressions? Reexamining the diagnostic value of appearance-based inferences. Journal of Experimental Social Psychology 46(2), 315–24.
 Todorov, Alexander, Mandisodza Anesu, Goren Amir & Crystal C. Hall (2005). Inferences of competence from faces predict election outcomes. Science 308(5728), 1623–1626.
 Chen, Fang Fang, Yiming Jing & Jeong Min Lee (2005). The looks of a leader: Competent and trustworthy, but not dominant. Journal of Experimental Social Psychology 51, 27-33.
 Mazur, Allan, Julie Mazur & Caroline Keating (1984). Military rank attainment of a West Point class: Effects of cadets' physical features. American Journal of Sociology 90(1), 125-150.
 Graham, John R., Campbell R. Harvey & Manju Puri (2017). A corporate beauty contest. Management Science 63(9), 3044–3056.
 Rezlescu, Constantin, Brad Duchaine, Christopher Y. Olivola & Nick Chater (2012). Unfakeable facial configurations affect strategic choices in trust games with or without information about past behavior. PLoS ONE 7(3), e34293.
 Porter, Stephen, Leanne ten Brinke & Chantal Gustaw (2010). Dangerous decisions: The impact of first impressions of trustworthiness on the evaluation of legal evidence and defendant culpability. Psychology, Crime & Law 16(6), 477–491.
 Mattarozzi, Katia, Valentina Colonnello, Francesco De Gioia & Alexander Todorov (2017). I care, even after the first impression: Facial appearance-based evaluations in healthcare context. Social Science & Medicine 182, 68–72.
Gaëtan Béghin, M. Ps., Ph.D. / D.Psy candidate, Laboratoire des Processus de Raisonnement, Université du Québec à Montréal. Translated by Susan D. Renaud.
How to cite this entry
Béghin, G. (2021). First impressions from faces, trans. S. D. Renaud. In E. Gagnon-St-Pierre, C. Gratton, & E. Muszynski (Eds). Shortcuts: A handy guide to cognitive biases. Online: www.shortcogs.com