DongWon Oh is a social psychologist interested in how people form impressions of others based on their looks and how new experiences further shape these initial impressions. To study this process, he focuses on the role of social biases and individual differences. He uses behavioral experiments, data-driven modeling and other reverse-correlational techniques, and neuroimaging.

Oh is currently a postdoc in Social Cognitive and Neural Sciences Lab at New York University (advisor: Jonathan Freeman). He has earned a PhD in Psychology at Princeton University (advisor: Alexander Todorov).

One line of my work is focused on understanding various perceiver factors that bias people’s initial social perception of others. These factors include people’s general biases about gender and race as well as the perceiver’s own social environment. These biases influence one’s perception from the moment we see a person’s face. This line of research shows that social perceptions occur in the rich context of the perceiver’s beliefs and knowledge, including beliefs and knowledge learned from their specific environment (e.g., country). Theoretical integration between the way we think about stereotypes, personality, and face impressions – topics that traditionally have been studied independently – is clearly needed. My research aims to uncover these deeply intertwined relationships to provide a more complete account of social perception.

To investigate the various perceiver factors affecting intuitive judgments of others (e.g., the perceiver’s beliefs about other people, the perceiver’s unique experience), I employ various techniques, including statistical modeling, dimensionality reduction, experiments, multivariate pattern analysis, and representational similarity analysis. In my research, I have shown the biasing role of visual cues of (1) gender, (2) economic status, and  (3) the social environment in social perception. 


Gender cues in faces bias judgments of other individuals’ traits from faces. My colleagues and I built a statistical face model (Oh et al., 2019 Vis Res; Todorov & Oh, 2021 Adv Exp Soc Psychol) representing intuitive judgments of competence while controlling for facial attractiveness, a factor that confounds judgments of competence (Oh et al., 2019 Psychol Sci). Without the “halo effect” of attractiveness, it was masculinity cues that strengthened the impressions of competence. This finding reveals the underlying gender biases in competence impressions. More generally, I have found that people form more simplified impressions of women than of men (Oh et al., 2019 J Exp Psychol: Gen). Impressions of women are more highly intercorrelated than those of men (e.g., there is a stronger negative correlation between likability and dominance impressions for women vs. men). Further, the simplification of impressions was more severe for those who strongly believed gender stereotypes. This is consistent with the body of literature on benevolent sexism (i.e., women are likely to be punished for not conforming to gender norms), while demonstrating that counterstereotypical traits suggested at the level of faces receive backlash. At the same time, these findings show that it is essential to consider stereotypes in social facial perception.

People have more simplified impressions of women than of men. In this plot, the level of impression simplification is indexed by the level of all pairwise correlations across specific trait impressions (e.g., impression of likeability). Each dot represents the level of correlations between a pair of impression ratings (e.g., correlation betwen the impression of likeability and the impression of dominance across all participants). (from Oh et al., 2019 J Exp Psychol: Gen)

Cues about economic status bias competence judgments. The association between status and competence is common and strong across many cultures. By pairing clothes that appeared expensive or clothes that appeared less expensive with the same faces, we manipulated the ostensible economic status of individuals. Individuals were perceived as more competent when their faces were paired with “expensive” clothes (Oh et al., 2019 Nat Hum Behav). We then employed various measures to suppress the economic-status effect, including a brief presentation duration of the person images (about 0.1 sec), instructions to ignore the clothes, and the promise of a substantial reward (100 US dollars) for the most accurate response that ignored the clothes. The economic-status effect persisted despite all countermeasures. This is consistent with previous findings showing the strong mental association between the competence and status of other individuals. At the same time, these findings provide a novel demonstration of the difficulty of reversing widely shared mental associations in perception (in this case, an association between competence and economic status). This reality challenges researchers to devise new measures to intervene with such biases.


Pairticipants across studies rated faces paired with “richer” clothes as more competent than the same faces paired with “poorer” clothes. These participants include in-person Princeton University sample, older adults sample recruited in a mall in New Jersey, and online participants that were more diverse in age and other demographics.
a: Pairticipants were given various measures that discouraged them to rely on the clothes while judging competence of other individuals. x=mean competence rating of a face with it is paird with “poorer” clothes, y=mean competence rating of a face with it is paird with “richer” clothes; each dot = each face.
b: Participants were promised $100 to “accurately” judge how competent each person looked (that is, completely ignore the effect of the clothes). Strikingly, this did not change the biases induced by economic status cues in the clothes. x = each face; y = mean comeptence rating; each color = economic status of clothes paired with the face.(from Oh et al., 2019 Nat Hum Behav)


The perceiver’s social environment (e.g., other people around you) biases trait judgments from faces. My research found that individual perceivers’ beliefs about personality traits predicted how they judged other individuals’ traits from faces (Oh, Martin, & Freeman, invited revision). For example, when a perceiver believed that other individuals that are extraverted are also agreeable, the same perceiver tended to judge a face appearing to be more extraverted as more agreeable (and vice versa). Notably, the effect was found at the regional level as well, across over 40 countries. In a country where people believed that they were simultaneously extraverted and agreeable, perceivers in the country tended to judge a face as appearing more extraverted as more agreeable. These findings suggest that people may facially perceive others’ personality based on the actual personality of people they encounter in their environment. This is consistent with previous findings showing that when making intuitive judgments, people use information acquired through social learning, highlighting the culture and personal experiences as an origin of facial impressions.

One line of my work is focused on understanding continuous update in social perception, with information provided by new experiences. This update allows people to adapt to the ever-changing environment. This line of research shows that impressions of other individuals often drastically yet reliably change after the initial judgments, affecting various aspects of our perception and memory, including the facial identity perception and personality impressions. My research aims to identify the underlying forces behind people’s evolving perception of others.

Although superficial information such as facial features and social biases contributes to initial social judgments, various factors color the initial input at later stages. To dissect the complex, multi-staged process of social impression formation, I use techniques such as mouse-tracking, reverse correlation, continuous rating, and multivariate pattern analysis. In my research, I have identified the role of knowledge in different stages of dynamic impression updating, ranging from facial identity recognition to trait judgments about other individuals.

Newly learned person knowledge shapes the way we recognize faces (Oh et al. 2021 Cognition). My colleagues and I first found that when a perceiver believed that two famous individuals’ personalities were similar, their faces were perceived to be similar, above and beyond their actual physical facial similarity, as revealed by hand-motion trajectories in face-to-name and face-to-face perceptual-match tasks. Correspondingly, when a perceiver believed that two individuals were similar, the perceptual representations of their faces were similar, as revealed by a reverse-correlational technique. 

Moreover, the effect persisted when perceivers learned about novel faces randomly linked to specific personality traits (e.g., trustworthiness): When any two individuals were presented as having more similar personalities – independent of their physical facial similarity – their faces were perceived as more similar. These findings suggest that the link between person knowledge and facial identity perception is causal in nature, providing evidence that even facial individuation – a seemingly rudimentary perceptual process – is influenced by novel social knowledge. At the same time, these findings highlight the need to consider how new information is integrated after the initial judgments, so as to gain a more complete understanding of social impressions.

People’s impressions of others’ personalities can be reliably tracked over time and systemically explained. My colleagues and I are working to identify the computational and neural basis of real-time social perception. To collect data, participants in an MRI scanner, without an explicit task, viewed a film in which various characters naturally interacted with one another. A separate group of participants watched the same video and continuously rated a trait of one of the characters (e.g., friendliness) in real time. Preliminary analyses indicated that (1) the perceivers considerably changed their impressions of each character as they learned more about these individuals during the continuous real-time trait rating; (2) nonetheless, there was a high level of consensus across perceivers in how they changed their impressions of the character over time; and (3) the impressions of each character were tracked in multiple brain regions. These findings suggest that we need to consider both how impressions of others are initially formed and how new information about them is continuously integrated over time. A novel framework is needed to consider the meaningful real-time changes in impressions.

As a behavioral researcher who started his career as a clinical psychologist, I’ve always been fascinated by individual differences in various faculties of the human mind. In my current research, I focus on the impact of the perceiver factors (e.g., stereotypes) and the impact of new information (e.g., new experiences) on social perception. A stronger emphasis on these conceptual (vs. perceptual) components of social processing naturally pushes me to investigate individual differences; the level and types of stereotypes and information gathered via life experiences vary across perceivers (whereas perceptual information from others’ faces for example, is largely identical across all perceivers). Investigating individual differences is scientifically particularly valuable because it often sheds a light on how a psychological phenomenon works. In my research, I have identified several domains of social perception in which individual differences substantially affect the way we perceive others: Individual differences in (1) gender stereotypes, (2) age, and (3) personality preferences.


Stronger stereotypes simplify your impressions of others from faces. My colleagues and I have found that people form more simplified (less differentiated) impressions of women than of men (Oh et al., 2019 Psychol Sci; Oh et al., 2019 J Exp Psychol: Gen). Impressions of women are more highly intercorrelated than those of men; people believed that on average for example, impressions of likability and dominance for women were more strongly negatively related to each other, compared to those of vs. men, in people’s minds.

The simplification of impressions (or lower level of differentiations across impressions) was more severe for those who strongly believed gender stereotypes, both for male and female faces. For example, if you are a person who strongly believes that ‘women are on average emotional, nurturing, etc. than men and that men are on average competitive, aggressive, etc. than women’, you are more likely to form a more simplified impressions of women and men than those who hold weaker gender stereotypes. Sexism at the level of faces (i.e., individuals whose look do not conform to gender norms are punished, economically and socially) is harsher, when the perceiver has stronger gender stereotypes. These findings show that it is essential to consider individual differences in the role of stereotypes in social perception.

My colleagues and I previously found that individuals were perceived as more competent when their faces were paired with “expensive” clothes (Oh et al., 2019 Nat Hum Behav). Despite the subtlety in our manipulation, the effect was hard to control, surviving various countermeasures we coined (e.g., instructions to ignore the clothes, and the promise of a reward (100 US dollars) for the most “accurate” response that ignored the clothes).

In addition to this main finding, we also found the effects of participant age. We found two things: (1) Regardless of the types of attire paired with the faces (“richer” or “poorer”), older participants reported higher competence impressions (i.e., rated other individuals more favorably) than did younger participants; (2) interestingly, older participants showed smaller differences between their competence ratings of “richer” vs. “poorer” images than younger participants did. Put differently, older participants were more generous in competence judgments; and not only that, older participants were more “immune” to the status-cue effects in competence impressions than younger participants.

In an industrialized society (like the US), women find male faces with feminine shape (e.g., bigger eyes, more narrow chin) compared to faces with masculine shapes. My colleagues and I found that this may be because women’s prefer feminine (vs. masculine) personality in a partner: warmth, nurturingness, and gentleness (Oh et al., 2020 J Exp Pscyhol: Hum Percept Perform).

In addition to this general principle (feminine personality preference → feminine face-shape preference), we also found a finding specific to individual perceivers: Those who prefer feminine personality traits in a partner prefer feminine characteristics in male faces, whereas those who prefer masculine personality traits (even thought they were in minority) prefer masculine characteristics in male faces. This finding explains the rich diversity in attractiveness perception and mating choice. At the same time, it highlights that we should consider conceptual factors when we think about interpersonal attraction, in addition to perceptual factors (e.g., facial symmetry, which contributes to attractiveness).