The More We Know about Others, the More We Think They Know about Us
Acquiring even mundane details about a stranger can shake our sense of anonymity.The More We Know about Others, the More We Think They Know about Us
Many entrepreneurs, particularly female ones, have horror stories about their pitch sessions with potential investors. Chicago Booth’s Waverly Deutsch recalls when two female cofounders were promoting their idea of a consumer marketplace business and the potential investor asked if they were a couple—and indicated he would be more interested in the company if they were. “Another huge problem is women building companies that serve women and being told, ‘Oh, I don’t know anything about that market, so I can’t invest in it,’ or ‘Why are you ruling out men as a market?’ or ‘I asked my wife about your business, and she wasn’t excited,’” says Deutsch, who has interviewed scores of entrepreneurs and written about their challenges in several columns for Chicago Booth Review. From the moment women arrive to introduce their ideas, forces seem to be working against them and undermining their credibility.
Booth’s Alexander Todorov has documented the deep biases that shape those first impressions. In a series of research projects, he has found that even before people open their mouths, their faces prompt automatic judgments and impressions laden with gender stereotypes. In three studies, Todorov and his collaborators adjusted the masculinity and femininity of both men’s and women’s faces and tested how people judge those faces on traits including dominance, competency, trustworthiness, and attractiveness. All of the findings suggest that we make quick assumptions steeped in biases. These snap judgments are formed with superficial cues, Todorov says, often painting a false picture of the person. In the process, all of us tend to overestimate some people and underestimate others, often women.
Addressing these biases could have a powerful economic impact. Supporting more women entrepreneurs would boost the global economy by nearly $5 trillion, according to the Boston Consulting Group. If there were a more equal distribution of men and women running businesses, global GDP would rise by about 3–6 percent, BCG suggests.
First impressions are not the only cause of inequality in investment, but Deutsch has seen their effect many times working with entrepreneurs. She has written about how knowing gender sets an investor up to ask different questions. (For more, read “What venture capitalists can learn from racist rats.”) And how investors react to facial expressions, tone of voice, and inflection, particularly in the first five seconds of a pitch. (Read “Entrepreneurs, remember the power of a smile.”) Todorov’s research suggests the impact of such initial impressions is widespread.
Facial features play a part in how competent we assume people to be, especially when we have no other information to go on.
When Susan Boyle first took the stage at Britain’s Got Talent in 2009 and confidently announced her desire to become a famous, professional singer, she was met with eye rolls, side-eyes, and smirks from the judges and audience alike. No one expected the 47-year-old woman with a protruding brow bone, double chin, and frumpy dress to have a beautiful voice. But Boyle shocked the audience (and later the world) when she sang a rendition of “I Dreamed A Dream” from the musical Les Misérables.
Social cues—including her age, quirkiness, and low socioeconomic class—played a role in viewers’ gut reaction to her, but research suggests that her more stereotypically masculine facial features may also have led people to underestimate her.
A study by DongWon Oh, Ron Dotsch, Jenny Porter, and Todorov—conducted when all were at Princeton—examined how dominant facial characteristics shape people’s impressions of women compared with men. Across three experiments, the researchers find that people use the same visual information when assessing traits such as trustworthiness and dominance from both men and women’s faces, but they evaluate that information differently on the basis of gender.
The researchers asked study participants to rate different sets of female and male faces—all white in some experiments, and a mix of different races in others—on traits including aggression, confidence, dominance, trustworthiness, intelligence, and caring. These experiments established that people, regardless of their own gender, use the same visual cues to judge both male and female faces. Boyle’s large brow bone and strong jaw, for example, lend her a more masculine appearance, which the research suggests would lead people to assume she has more masculine character traits.
In further experiments, the researchers used the earlier results to manipulate male and female faces to look more or less trustworthy and dominant, and then asked survey takers to rate those faces along those traits. Participants were always told to rely on their “gut feeling” when rating each face.
Overall, the findings provide evidence for the backlash effect, an established phenomenon in which women who defy expected stereotypes of femininity experience social or economic penalties. Results of the experiments reveal a difference in how perceived masculinity and femininity affected participants’ impressions of men and women. Women who were deemed to have more masculine traits and features were evaluated more negatively than men deemed to have feminine features. In short, women whose appearance countered what participants expected were judged more harshly than men whose appearance defied expectation.
For women, looking different than people expect tends to garner negative judgment, as played out when Susan Boyle stepped on stage. Boyle, who made people think twice about those stereotypes, finished second in her season of Britain’s Got Talent and ultimately realized her ambition of becoming a professional singer. After signing to the label run by show host Simon Cowell, she released eight albums and, as of February 2021, had reportedly sold 19 million records.
To study snap judgments and uncover issues such as gender bias, researchers may show people a series of faces—and in many cases, those images are computer generated and carefully manipulated.
Chicago Booth’s Alexander Todorov might give a slight smile to a neutral face or increase the distance between someone’s eyebrows and eyes to study what features are considered, say, friendly or trustworthy. But how slight should the smile be? Or how much space should be added under an eyebrow? There could be millions of variations, if not more. To address this, Dartmouth postdoctoral scholar Nikolaas N. Oosterhof and Todorov developed computational, data-driven models of social judgments of faces. They have used their models to create a number of databases that they make available to other researchers.
Their work in this area builds on that of Volker Blanz and Thomas Vetter, both of the Max Planck Institute for Biological Cybernetics in Germany. Blanz and Vetter created a way to essentially represent each face as a set of numbers. Any face can map onto this “face space.”
Oosterhof and Todorov, in turn, randomly generated sample faces and asked study participants to rate those faces on a particular characteristic, such as trustworthiness. Using these ratings, they could model the specific social judgment (trustworthiness) as a function of the variation in face shape and face “reflectance,” which they define as the brightness, texture, and color variation of a face.
Their research reduces differences in faces to 100 principal components (PCs)—50 for face shape and 50 for reflectance. Each PC is responsible for increasingly smaller variations. For example, when it comes to face shape, PC1 defines the overall width of a face and all of its features. PC2 relates to the elongation of the face. Lower PCs change the face shape in broad sweeps, higher ones in subtler ways.
A similar 50-dimensional model represents differences in face reflectance, which Todorov says is just as important to our perception of a person as face shape. PC1 again has the greatest effect, determining the overall lightness or darkness of a face.
The cumulative statistical model can be used to essentially map any face along its coordinates. The researchers can also adjust PCs to slightly tweak a face’s appearance. Their approach allows them to generate an infinite number of faces and parametrically manipulate them according to the social judgment they’re evaluating.
Oosterhof and Todorov, as well as Todorov and New York University postdoctoral scholar DongWon Oh, have used these frameworks of social judgments of faces to generate and validate many databases that serve as models of perceived trustworthiness, dominance, attractiveness, competence, extroversion, likability, and more. As of April 2020, more than 4,380 researchers had downloaded these face databases.
Women perceived as too masculine can be judged harshly, and yet some masculinity broadcasts confidence.
Fashion illustrates this through the shoulder pad. It was a key part of 1980s power dressing, in which women office workers dressed more like their male colleagues. “Shoulder pads add to an authoritative appearance, which can be incredibly useful for women,” says University of Nevada, Las Vegas’s Deirdre Clemente, who specializes in the history of clothing and fashion.
The goal of shoulder pads was to make women look bigger, broader, more competent. Although they may not have had the science to back it up, women understood that if they wanted to work with and be taken seriously by men, they had to look at least a little more masculine.
A study from Oh, now a postdoctoral scholar at New York University, Princeton’s Elinor Buck, and Todorov looked at what the researchers call “the visual ingredients of the competence stereotype.” Facial features play a part in how competent we assume people to be, especially when we have no other information to go on. Over four experiments, Oh, Buck, and Todorov identified three main components of competence impressions: attractiveness, confidence, and masculinity.
In some experiments, the researchers asked online survey takers to rate faces on the basis of how competent or how attractive they thought the person to be. In others, they asked participants to rate faces as either male or female on the basis of confidence and masculinity. The faces were computer generated and had no markers, such as hair or clothing, to indicate gender. In a final experiment, the researchers used photorealistic faces made to look more true to life than the previous images. They showed survey takers faces of one gender (with the markers included this time) and asked them to judge how competent the faces seemed to be. Again, each participant was told to rely on gut instinct.
Taken together, the studies find that the most competent-looking faces were those that were perceived as more confident and masculine. In the experiments in which the gender of each face was not specified, raters tended to categorize the faces they deemed competent as male rather than female.
Although perceived attractiveness plays a part in whether or not people judge someone competent, and feminine facial features are typically considered more attractive on both women and men, the researchers find that perceived masculinity is an important and strong component as well. This may explain the shoulder-pad trend when bias against women was more explicitly rampant in workplaces. The fundamental purpose of shoulder pads is to distort the body, Clemente says. “Shoulder pads were part of a broader aesthetic change, and a lot of this was to give this power and authority and to make somebody’s body look larger and more powerful than it actually was,” she says. Women did what they could to make themselves look more masculine, knowing at least subconsciously that a masculine appearance mattered. Shoulder pads, Clemente says, are again becoming popular.
However, Todorov and his colleagues note that there’s a fine line, as is demonstrated with the backlash effect. Some masculinity may help women be judged as more competent. Yet too much masculinity can lead to discrimination.
Some argue that blind auditions haven’t gone far enough, but Todorov notes they have successfully countered faulty first impressions.
Gender stereotypes can affect men, too, but often in a positive way. When women such as Susan Boyle display masculine traits, they tend to be judged negatively for it when men who display feminine traits are not. Indeed, in the experiments by Oh, Buck, and Todorov, when men displayed masculine traits, they were assumed to be more competent. It can even work to their advantage.
In the realm of snap judgments, one thing people quickly size up is attractiveness, a quality that carries weight both socially and professionally. People expect someone who is attractive to do well not just on dates but in the workplace and life more generally. A host of studies and models of facial impressions, dating back decades, supports the idea that someone considered attractive is perceived to be more competent and to have a higher social status. And according to a study from Oh, Columbia’s Natalie Grant-Villegas, and Todorov, men with more feminine facial features are considered particularly attractive.
The study explored how facial femininity and masculinity correlated with perceived character traits. The researchers asked two sets of heterosexual women to rate a set of white male faces on six stereotypically masculine and feminine personality traits: warmth, nurturance, gentleness, dominance, confidence, and competitiveness. Overall, the women indicated that men with some feminine facial features, including large eyes and a softer jawline, were warmer, more nurturing, and gentler.
After rating faces on personality traits, both sets of women were asked to rate 75 male faces on attractiveness. The faces were manipulated to look more or less masculine. In both studies, straight women more often rated the more feminine faces as attractive.
The researchers are careful to point out that this isn’t universally true. Some straight women may prefer more masculine personalities and therefore be more attracted to masculine faces. Generally the straight women who participated in the study preferred the faces they deemed warmer, more nurturing, and gentler, but how attractive you find someone is highly personal. There’s also no rule of nature that says a man who looks masculine isn’t also gentle and nurturing. Such perceived character traits are steeped in gendered stereotypes.
That said, overall, stereotypes seem to help men either way. If they’re masculine, they’re considered competent. If they’re feminine, they’re attractive—likely loving, caring partners or good-looking people who will do well in life. Yet women are doubly hurt by snap judgments—too feminine to be taken seriously, until they’re judged to look so masculine they trigger the backlash effect.
Knowing that we’re prone to often incorrect and biased impressions isn’t enough to keep them from happening. “It’s difficult to restrain from engaging in snap judgments,” Todorov says. “They are fairly automatic, but superficial.”
People can judge another person’s face after seeing it for only 100 milliseconds, he finds in a study run with Janine Willis, a Princeton student at the time and now a corporate counselor. That’s 0.1 seconds that our brains take to process a face and make a choice about whether that person is trustworthy, attractive, or dominant.
Willis and Todorov showed 117 undergraduates photos of faces for three different time periods: one-tenth of a second, half a second, and 1 second. In one experiment, participants were asked to rate the faces on trustworthiness, with each subsequent experiment testing a different characteristic: attractiveness, likability, competence, and aggressiveness.
The participants did not need the extra time: the judgments they made about each face were not significantly different when they were given longer. This suggests that our appearance-based judgments are immediate and difficult to change, the researchers conclude. “The best course of action is to have access to good quality information and make sure that this information dominates your decision,” Todorov says.
He gives the example of symphony orchestras. For decades, large symphony orchestras in the United States consisted almost entirely of white men, and their conductors largely controlled the hiring process. Recognizing the process needed to be fairer and more open, many of the biggest orchestras in the country adopted blind auditions in the 1970s and ’80s. From then on, a musician’s interview process included playing a piece for a group of judges behind a screen, so the judges did not know the performer’s gender or race. The new procedure worked well for women. In 2000, Harvard’s Claudia Goldin and Princeton’s Cecilia Rouse examined the impact of blind auditions, finding that a blind process increased the probability that a woman musician would advance to the next audition stage and had an even bigger impact on the likelihood that she would be hired.
Some argue that blind auditions haven’t gone far enough, but Todorov notes they have successfully countered faulty first impressions. “One of the implications of my research is that to overcome biases, you would need to avoid presenting any cues triggering the bias,” says Todorov. Cues can include even the click of high heels as the person wearing them walks across a stage.
The same idea can be applied to other situations, even start-up investing. If investors were to narrow the pool of ideas without first seeing the faces of the people at the helm, women could potentially get a greater share of venture capital.
Deutsch wrote in CBR about VC firms that are using artificial-intelligence engines to help them vet and guide investments. Connetic Ventures is one such firm, and using A.I., it produced a diverse portfolio, with just over a third of companies led by female CEOs. (Read “Women and minority investors are taking matters into their own hands.”) Seemingly objective criteria may not eliminate bias entirely, and A.I. can develop its own biases by finding variables that correlate to gender. But in this case, Todorov notes, automating the process would seem to have helped remove the influence of snap judgments. That allows investors to react to ideas, not facial features.
Acquiring even mundane details about a stranger can shake our sense of anonymity.The More We Know about Others, the More We Think They Know about Us
The pursuit of "mimetic dominance" over others can raise people's valuation of a product above their own intrinsic taste for it.The Equation: How to Put a Price on People’s Appetite for Exclusivity
Chicago Booth’s John Paul Rollert reviews the ups and downs of disruption.What’s the Matter with Disruption?
We want to demonstrate our commitment to your privacy. Please review Chicago Booth's privacy notice, which provides information explaining how and why we collect particular information when you visit our website.