Subliminal Priming Reveals Two Faces of Facial Expressions

Facial emotional expressions can serve both as emotional stimuli and as communicative signals. The research reported here was conducted to illustrate how responses to both roles of facial emotional expressions unfold over time. As an emotion elicitor, a facial emotional expression (e.g., a disgusted face) activates a response that is similar to responses to other emotional stimuli of the same valence (e.g., a dirty, nonflushed toilet). As an emotion messenger, the same facial expression (e.g., a disgusted face) serves as a communicative signal by also activating the knowledge that the sender is experiencing a specific emotion (e.g., the sender feels disgusted). By varying the duration of exposure to disgusted, fearful, angry, and neutral faces in two subliminal-priming studies, we demonstrated that responses to faces as emotion elicitors occur prior to responses to faces as emotion messengers, and that both types of responses may unfold unconsciously. Facial emotional expressions have the capacity to evoke all kinds of responses. After all, ‘‘the face has the only skeletal muscles of the body that are used, not to move ourselves, but to move others’’ (Smith & Scott, 1997, p. 229). Perceivers of a facial emotional expression may not only respond to the expression itself, but may also react to the communicative message it reveals. A happy face, for example, may act as an emotion elicitor and elicit positive affect because, just like a picture of a young puppy, a happy face is a positive stimulus. However, such a face may also act as an emotion messenger and reveal specific knowledge about the motives and intentions of the sender. The happy face of a salesperson, for example, may reveal his or her ambition to sell a product. Both roles of facial emotional expressions seem crucial in human social functioning. That is why we propose that responses to both roles may be activated efficiently and unconsciously. An essential difference between the two types of responses, however, is their specificity: Responses to the emotion-elicitor role of facial emotional expressions often are global and valence based because they are based on an initial assessment of the stimulus, whereas responses to the emotion-messenger role are relatively specific and knowledge based because they are based on secondary and relatively detailed information processing. These characteristics strongly suggest that during impression formation, responses to the emotion-elicitor role may occur earlier than responses to the emotion-messenger role. To test this dual-role perspective on automatic responses to facial emotional expressions, we conducted two subliminal-priming studies investigating the timing of emotion-elicitor and emotion-messenger effects of facial emotional expressions. FACIAL EXPRESSIONS AS EMOTION ELICITORS AND MESSENGERS Facial emotional expressions are frequently referred to as the key in understanding human emotions (e.g., Russell & FernandezDols, 1997). Research has revealed that facial emotional expressions can have different functions (Hess, Philippot, & Blairy, 1998; Keltner & Haidt, 1999). When they serve the role of emotion elicitor, they resemble any other emotional stimulus (see Murphy & Zajonc, 1993; Ruys, Spears, Gordijn, & De Vries, 2007; Stapel, Koomen, & Ruys, 2002; Winkielman, Berridge, & Wilbarger, 2005). A disgusted facial expression, for example, is similar to a dirty toilet bowl in that both stimuli may primarily elicit a global, negative withdrawal reaction. As emotion elicitors, facial emotional expressions may activate global, valencebased responses that may influence all kinds of unrelated judgments, preferences, and behaviors of the perceiver. Address correspondence to Kirsten I. Ruys, Tilburg Institute for Behavioral Economics Research (TIBER) at Tilburg University, P.O. Box 90153, 5000 LE Tilburg, The Netherlands, e-mail: [email protected]. PSYCHOLOGICAL SCIENCE Volume 19—Number 6 593 Copyright r 2008 Association for Psychological Science Facial expressions also have a different function, namely, that of emotion messenger (see Ekman, 1992; Fridlund, 1994; Izard, 1994; Jakobs, Fischer, & Manstead, 1997; Keltner & Haidt, 1999; Russell, 1994). As emotion messengers, facial emotional expressions signal information about the sender’s feelings or motivations that is essential for successful communication. An angry facial expression, for instance, may signify the expresser’s discomfort and dissatisfaction with a person or situation. To date, most research regarding the emotion-messenger role of facial expressions has focused on the first stages of emotion communication, that is, the recognition of facial emotional expressions and the meaning perceivers attach to these expressions (e.g., Ekman, 1992; Fernandez-Dols & Carroll, 1997). However, what seems crucial in the emotion-messenger role of facial emotional expressions is that perceivers not only detect the emotion (as in the case of the emotion-elicitor role), but also detect the senderemotion link and attribute the emotion to the sender (Adolphs, 2006; Dimberg & Öhman, 1996). Thus, a response to the emotionmessenger role of, for example, an angry-looking boss, is a response to ‘‘my boss is angry,’’ whereas a response to the emotionelicitor role of this same stimulus is a response to anger in general. UNCONSCIOUS ELICITATION Despite this important difference in specificity between reactions to the emotion-elicitor role and responses to the emotion-messenger role, we think both types of responses may occur quickly and without conscious awareness. The quick and effective evaluative processing of facial emotional expressions can help people to immediately escape from danger and to take advantage of potential opportunities. Thus, unconsciously detecting the friendly or aggressive facial expression of an opponent can be important for one’s survival (Atkinson & Adolphs, 2005; Cacioppo & Gardner, 1999; Zajonc, 1980). This view is supported by empirical evidence showing that subliminal exposure to facial emotional expressions elicits emotion-congruent physiological reactions and brain-activation patterns (De Gelder, 2005; Dimberg, Thunberg, & Elmehed, 2000; Phillips et al., 2004) and may unconsciously color unrelated preferences and judgments (Murphy & Zajonc, 1993; Winkielman et al., 2005). This evidence regarding the unconscious detection of facial emotional expressions pertains mostly to their emotion-elicitor role. More recent research, however, suggests that an unconscious reaction to their emotion-messenger role—requiring the activation of more specific knowledge—is also possible. This work shows that subliminally presented facial emotional expressions can evoke not only global, valence-based affective reactions, but also more specific, knowledge-based affective reactions, such as affective responses that are sensitive to facial gender or ethnicity information (Ruys et al., 2007; Stapel & Koomen, 2006; Stapel et al., 2002). The important point for present purposes is that people are capable of unconsciously detecting descriptive, knowledgebased information (i.e., the social category of the face), in addition to global, valence-based information (i.e., the expressed emotion). It thus seems a likely possibility that responses to both the emotion-elicitor role and the emotion-messenger role unfold unconsciously. ADDITIONAL PROCESSING OF THE SENDEREMOTION LINK What is less clear, however, is when the emotion-elicitor and emotion-messenger functions come into play. This information is crucial because responses to the emotion-elicitor and emotionmessenger roles of facial emotional expressions differ and may sometimes even be opposite (see Lanzetta & Englis, 1989). For example, seeing a smile may evoke a general positive response (to the role of the smile as an emotion elicitor), but when this smile comes from one’s worst enemy, it may evoke a more specific, probably quite negative reaction (to the role of the smile as an emotion messenger), such as anger. How can researchers separate these two types of responses? We think the answer to this question lies in how they unfold over time: Global evaluative (emotionelicitor) reactions to stimuli are typically triggered earlier than specific descriptive (emotion-messenger) reactions. This is because information processing often unfolds from the global to the local (e.g., Rosenthal, 2004; Werner, 1956). Neurological research supports this view, showing that there are independent systems for coarse, evaluative processing and detailed, perceptual processing (e.g., Adolphs, 2003; LeDoux, 1989). Moreover, research on event-related potential (ERP) brain responses to emotional faces (Palermo & Rhodes, 2007) has shown that crude affective categorization often occurs rapidly, whereas fine-grained processes necessary to recognize the identity of a face or to discriminate between basic emotional expressions typically need more time. Thus, responses to the emotion-messenger role may need more time to develop than responses to the emotion-elicitor role because the former require the activation of knowledge (i.e., the sender-emotion link) in addition to the activation of the emotion (Atkinson & Adolphs, 2005). The divergent temporal characteristics of responses to the elicitor versus messenger role allow one to separate these responses by varying the duration of exposure to facial emotional expressions. As we showed in earlier research using parafoveal subliminal priming, super-quick subliminal exposures to priming stimuli trigger primarily global, valence-based responses, whereas quick subliminal exposures to priming stimuli trigger mostly specific, knowledge-based responses (see Ruys & Stapel, 2008; Stapel & Koomen, 2005, 2006; Stapel et al., 2002). For example, super-quick subliminal priming of concepts (e.g., honest vs. dishonest) elicited global, valence-based reactions (positive vs. negative), whereas quick subliminal priming of these concepts elicited knowledge-based reactions (honest vs. dishonest). 594 Volume 19—Number 6 Emotion Elicitor or Emotion Messenger?