Making Up Your Mind After a 100-Ms Exposure to a Face

People often draw trait inferences from the facial appearance of other people. We investigated the minimal conditions under which people make such inferences. In five experiments, each focusing on a specific trait judgment, wemanipulated the exposure time of unfamiliar faces. Judgments made after a 100-ms exposure correlated highly with judgments made in the absence of time constraints, suggesting that this exposure time was sufficient for participants to form an impression. In fact, for all judgments—attractiveness, likeability, trustworthiness, competence, and aggressiveness—increased exposure time did not significantly increase the correlations. When exposure time increased from 100 to 500 ms, participants’ judgments became more negative, response times for judgments decreased, and confidence in judgments increased. When exposure time increased from 500 to 1,000 ms, trait judgments and response times did not change significantly (with one exception), but confidence increased for some of the judgments; this result suggests that additional time may simply boost confidence in judgments. However, increased exposure time led to more differentiated person impressions. Lavater’s (1772/1880) Essays on Physiognomy, which was written in 1772 and reprinted inmore than 150 editions by 1940, described in minute detail how to relate facial features to personality traits (e.g., ‘‘the nearer the eyebrows are to the eyes, the more earnest, deep, and firm the character,’’ p. 59). Although these ideas strike most people today as ludicrous and bring to mind phrenology, empirical evidence shows that the effects of facial appearance on social outcomes are pervasive. In almost every significant domain of life, attractive people get better outcomes than unattractive people (Hamermesh & Biddle, 1994; Zebrowitz, 1999). The effects of baby-faced appearance are as pervasive as are the effects of attractiveness (Montepare& Zebrowitz, 1998; Zebrowitz, 1999). For example, baby-faced individuals are less likely to receive severe judicial outcomes than mature-faced individuals (Zebrowitz & McDonald, 1991). From the structure of the face, people form not only global impressions, but also specific trait impressions (Hassin& Trope, 2000). For example, we showed that inferences of competence, based solely on facial appearance, predicted the outcomes of U.S. congressional elections in 2000, 2002, and 2004 (Todorov, Mandisodza, Goren, & Hall, 2005). Although we measured impressions on a variety of traits, including attractiveness, trustworthiness, and likeability, the trait inference that predicted the election outcomes was competence. Competence was also rated as the most important attribute for a person running for a public office. This finding suggests that person attributes that are important for specific decisions are inferred from facial appearance and influence these decisions. From both the standard-intuition and the rational-actor points of view, trait inferences from facial appearance should not influence important deliberate decisions. However, to the extent that these inferences occur rapidly and effortlessly, their effects on decisions may be subtle and not subjectively recognized. Using the terms of dual-process theories (Chaiken & Trope, 1999; Kahneman, 2003), we have argued that trait inferences from faces can be characterized as fast, intuitive, unreflective System 1 processes that contrast with slow, effortful, and deliberate System 2 processes (Todorov et al., 2005). We provided preliminary evidence for this proposal by showing that inferences of competence based on 1-s exposure to the faces of the winners and the runners-up for the Senate races sufficed to predict the election outcomes. In this article, we report a series of studies in which we systematically manipulated the exposure time of faces to further explore the minimal conditions under which people make trait inferences from facial appearance. Research on visual processing has shown that high-level object representations can be constructed very rapidly from visual scenes (Grill-Spector & Kanwisher, 2005; Rousselet, Fabre-Thorpe, & Thorpe, 2002; Thorpe, Fize, & Marlot, 1996). It is possible that inferences about socially significant attributes are also rapidly extracted Address correspondence to Alexander Todorov, Department of Psychology, Green Hall, Princeton University, Princeton, NJ 085441010, e-mail: [email protected]. PSYCHOLOGICAL SCIENCE 592 Volume 17—Number 7 Copyright r 2006 Association for Psychological Science from facial appearance. We conducted five experiments, each focusing on a different judgment from facial appearance: attractiveness, likeability, competence, trustworthiness, and aggressiveness. Among the studied traits, attractiveness is the only one that is unambiguously related to facial appearance; that is, it is a property of facial appearance. In this respect, judgments of attractiveness provide a benchmark for judgments of character traits. Liking is a global affective response that may require minimal inferential activity (Zajonc, 1980). In contrast to attractiveness and liking, trustworthiness, competence, and aggressiveness are specific traits that have clear behavioral manifestations. These traits are also important for both social and economic interactions. In all the experiments, faces unfamiliar to the participants were presented for 100 ms, 500 ms, or 1,000 ms. For each face, participants were asked to make a trait judgment and then to express their confidence in that judgment. We tested three hypotheses: (a) that a 100-ms exposure to a face is sufficient for making a trait judgment, (b) that additional exposure time increases confidence in trait judgments without necessarily changing the judgments, and (c) that additional exposure time allows for more differentiated trait impressions. If trait inferences from faces can be characterized as System 1 inferences, minimal exposure time should be sufficient for trait inferences to occur. In order to obtain criterion judgments, we asked a large group of participants to make trait judgments of the faces in the absence of time constraints. If a 100-ms exposure to a face is sufficient for making a trait inference, then trait judgments made after 100-ms exposure should correlate with judgments made in the absence of time constraints. In contrast, if 100 ms is insufficient, these judgments should be uncorrelated, and only judgments made after longer exposures should correlate with judgments made in the absence of time constraints. We were also interested in how additional exposure time affects trait judgments and confidence in these judgments. If people commit to a judgment early in time, additional time can serve only as a justification of this judgment. If this is the case, confidence should increase as a function of exposure time, but there should be no corresponding change in judgment. For example, if 500-ms exposure is sufficient for participants to form stable trait judgments, little change in judgments should be observed with additional exposure time. However, additional exposure time may boost confidence in judgments. Even if trait impressions can be formed after minimal exposure time, additional time may allow for more differentiated impressions. For example, it is possible that after 100-ms exposure, people perform a coarse affective discrimination of faces, such that judgments of different traits are highly correlated. Additional time may allow for more fine-grained impressions based on specific trait attributions, in which case judgments of different traits would be less correlated. We tested these predictions using factor analysis. METHOD Participants A total of 245 undergraduate students from Princeton University participated in the studies either for payment or for partial course credit. One hundred twenty-eight participated in a preliminary study in which we obtained measures of trait inferences from facial appearance in the absence of time constraints. One hundred seventeen participated in the fivemain experiments; 20 were asked to make attractiveness judgments, 25 to make liking judgments, 23 to make competence judgments, 24 to make trustworthiness judgments, and 25 to make aggressiveness judgments. Stimuli In all the studies, we used a database of photographs of 70 amateur actors, 35 females and 35 males between 20 and 30 years of age (Lundqvist, Flykt, & Öhman, 1998). In the pictures, all actors wore gray T-shirts, and there were no beards, mustaches, earrings, eyeglasses, or visible makeup. We used frontal head-shot photographs of individuals with neutral expressions. Of the 70 photographs, 2 photographs of males were excluded because of poor quality; we also excluded 2 photographs of females in order to have equal numbers of male and female photographs. To obtain reliable measures of trait inferences from facial appearance, we presented participants in the preliminary study with the photographs and asked them to judge the degree to which the person in each picture was attractive, likeable, competent, honest or trustworthy, aggressive, extraverted or enthusiastic, sympathetic or warm, dependable or self-disciplined, calm or emotionally stable, open to new experiences or complex, and ambitious. The judgments on the first five dimensions provided the criterion judgments for the five experiments. In the preliminary study, each face was presented on a separate questionnaire page, and the order of the trait judgments was fixed. All judgments were made on a 9-point scale ranging from 1 (not at all) to 9 (extremely). The photographs were randomly divided into three groups, each one containing the same number of males and females, and for each group of photographs, we generated two random orders. Participants were randomly assigned to one of the six sets of photographs (3 groups 2 orders) and completed the task at their own pace. Each photograph was rated by 42 or 43 participants. The trait judgments were highly reliable. For the three groups of photographs, the Cronbach alphas were .97, .96, and .95 for attractiveness; .94, .91, and .89 for likeability; .92, .92, and .92 for trustworthiness; .85, .91, and .96 for competence; and .87, .75, and .89 for aggressiveness. The mean trait judgments across participants served as the criterion judgments for the experiments. To the extent that limited exposure time is sufficient for people to form trait impressions from faces, the experimental judgments made under Volume 17—Number 7 593 Janine Willis and Alexander Todorov