Biased Recognition of Happy Facial Expressions in Social Anxiety

Recognizing emotional expressions is central to understanding the feelings and intentions of other people. Little is known about how social anxiety affects the recognition of emotional expressions. Recent research finds a recognition advantage for happy expressions over negative expressions. In two experiments, social anxiety moderated the recognition advantage of happy faces. People low and high in social anxiety recognized sad faces (Experiment 1) and angry faces (Experiment 2) equally quickly, but people high in social anxiety took longer to recognize happy faces. Both groups showed a significant recognition advantage for happy faces, although the advantage was at least twice as large in the low social-anxiety group. The discussion focuses on mechanisms connecting social anxiety to face processing and on the role of expression recognition in other emotionalprocessing biases. Article: Recognizing emotional expressions is central to understanding the feelings and intentions of other people. Little is known about how social anxiety affects the recognition of emotional expressions. Recent research finds a recognition advantage for happy expressions over negative expressions. In two experiments, social anxiety moderated the recognition advantage of happy faces. People low and high in social anxiety recognized sad faces (Experiment 1) and angry faces (Experiment 2) equally quickly, but people high in social anxiety took longer to recognize happy faces. Both groups showed a significant recognition advantage for happy faces, although the advantage was at least twice as large in the low social-anxiety group. The discussion focuses on mechanisms connecting social anxiety to face processing and on the role of expression recognition in other emotionalprocessing biases. Social anxiety is characterized by persistent fears of interacting with other people, performing before others, and being observed, scrutinized, and evaluated (Clark & Wells, 1995; Leary & Kowalski, 1995). Social anxiety is viewed as a continuum, ranging from social fearlessness to subclinical social anxiety to clinical levels of social phobia (McNeil, 2001; Rapee & Heimberg, 1997). At subclinical levels, social anxiety is relatively common-approximately 20% of people report having irrational social fears (Furmark et al., 1999). Lifetime prevalence of clinical social anxiety is approximately 5% for men and 10% for women (Wittchen, Stein, & Kessler, 1999), although some estimates place the lifetime prevalence at around 13 to 16% (Kessler, McGonagle, & Zhao, 1994; Wacker, Müllejans, Klein, & Battegay, 1992). Cognitive theories of social anxiety contend that information-processing biases are central to the disorder (Clark et al., 2003; Clark & Wells, 1995; Foa, Franklin, & Kozak, 2001; Rapee & Heimberg, 1997; Turk, Lerner, Heimberg, & Rapee, 2001). Biases in attention, memory, and reasoning are thought to exacerbate and maintain social anxiety. Consistent with the claims of cognitive models, research has discovered many biases in information processing in clinical and subclinical samples, such as negative interpretations of ambiguous events, appraisals of social situations as dangerous, vigilance for threatening information, and excessive self This research was presented at the 2004 meeting of the Society of Southeastern Social Psychologists, Clemson, SC. focused attention (see Amir & Foa, 2001; Foa, Franklin, Perry, & Herbert, 1996; Heinrichs & Hofmann, 2001; Spurr & Stopa, 2002). Many of social anxiety's cognitive biases involve processing emotional expressions. Facial expressions of emotion are powerful sources of social information (Darwin, 1872/1998; Ekman, 1993; Izard, 1971)-positive and negative emotional expressions quickly convey social messages (Keltner & Ekman, 2000). During social interactions, people gain much of their information about the feelings and intentions of others through emotional expressions, such as the liking and approval expressed by happy faces, the hostility expressed by angry faces, and the rejection expressed by disgusted and contemptuous faces. In a recent review, Heinrichs and Hofmann (2001) noted a need for more research on how socially anxious people process faces, because positive and negative facial expressions are more ecologically valid signals than are more widely studied stimuli (e.g., positive and negative words). To date, a small literature has investigated biased processing of facial expressions in social anxiety. Socially anxious people (1) direct their attention to angry, threatening faces instead of happy or neutral faces (Mogg & Bradley, 2002; Mogg, Philippot, & Bradley, 2004); (2) detect threatening faces faster than happy faces in visual search tasks (Gilboa-Schechtman, Foa, & Amir, 1999); (3) avoid looking at emotional faces when nonsocial stimuli can be viewed instead (Mansell, Clark, Ehlers, & Chen, 1999); (4) display biased memory for faces displaying angry and happy emotions (D'Argembeau, Van der Linden, Etienne, & Comblain, 2003; Lundh & Öst, 1996); (5) do not report increased social efficacy following a conversation with a partner displaying positive emotions (e.g., maintaining eye contact, nodding), instead seemingly discounting the interaction (Wallace & Alden, 1997); and (6) may show biased descriptions of the emotions presented on emotional faces (Merckelbach, van Hout, van den Hout, & Mersch, 1989; Philippot & Douilliez, 2005). What do these emotional-processing biases have in common? One cognitive process-the recognition of emotional expressions-is involved in each of these biases. For example, to direct attention to an angry instead of a neutral face, one must first recognize the emotion on each face. To show biased recall for faces showing certain emotions, one must have first recognized the emotion on the face to encode it with the face's identity. When attention is directed away from emotional faces, the emotional expression must first be decoded. Although the recognition of emotional expressions subserves several well-known biases in social anxiety, little is known about how socially anxious people recognize emotional expressions. Only a handful of experiments have explored whether social anxiety affects this fundamental aspect of emotional processing. One experiment found that children with clinical social phobia took longer to recognize all emotional faces (Melfsen & Florin, 2002), relative to nonphobic children. It is unclear, however, whether this result reflected an across-the-board deficit in emotion recognition or simply a disruptive effect of anxiety on performance. Another experiment found that socially anxious adults had a negative response bias when classifying faces as neutral or negative (Winton, Clark, & Edelmann, 1995). Response times were not recorded, however, thus leaving open the question of whether socially anxious people recognized some expressions faster than others. These two experiments are important early steps, yet they have some limitations. First, in both experiments, the faces were presented for 60 ms and followed by a pattern mask. It takes up to 170 ms to form a detailed structural representation of a face (Adolphs, 2002), so responses to brief-and-masked presentations will be overly influenced by response biases (such as the negative response bias found by Winton et al., 1995). Second, neither experiment distinguished between types of negative expressions. Participants were asked to give the same response to different negative expressions (e.g., anger, fear, disgust, contempt, and sadness). Recent research, however, shows that people do not process all negative emotions alike; angry expressions are particularly central to face processing (Fox et al., 2000; Lundqvist, Esteves, & Ohman, 1999; Tipples, Atkinson, & Young, 2002). Thus, these experiments do not assess potential differences between negative emotions. RECOGNIZING EMOTIONAL EXPRESSIONS People recognize emotional expressions automatically, meaning that they identify a face's emotional expression quickly, unintentionally, and effortlessly (Bruce & Young, 1986; Haxby, Hoffman, & Gobbini, 2000; Stenberg, Wiking, & Dahl, 1998). Like many automatic processes, the process of recognizing an emotional expression is highly conceptual (see Adolphs, 2002). Emotions cannot be recognized solely by perceiving visual features of the face. The retrieval and application of conceptual knowledge of emotions is necessary for understanding the face's emotional meaning and labeling the expression (Adolphs, 2002). Upon forming a structural representation of a face (Bruce & Young, 1986), people must retrieve higher-order conceptual information to label the perceptual representation as a "happy face," "angry face," or "sad face." Understanding the central role of conceptual knowledge in expression recognition is pivotal to appreciating how biases in expression recognition can emerge. The present experiments take as their starting point recent research on the recognition of positive and negative facial expressions (Leppänen & Hietanen, 2004; Leppänen, Tenhunen, & Hietanen, 2003). Many experiments find a happy-face recognition advantage, in which people recognize happy faces faster than negative facial expressions (Feyereisen, Malet, & Martin, 1986; Kirita & Endo, 1995; Leppänen & Hietanen, 2003). This effect is not due to simple differences in low-level face features; significant happy-face advantages appear for schematic faces that control for an expression's featural complexity (Kirita & Endo, 1995; Leppänen & Hietanen, 2004, Experiment 2). Instead, it reflects differences in how quickly people can retrieve and apply conceptual knowledge to the perceptual representation of the face. For example, activating positive emotional information boosts the advantage for happy faces, whereas activating disgusting emotional information causes an advantage for disgusted faces (Leppänen & Hietanen, 2003, Experiment 2). Research on the happy-face recognition advantage suggests that social anxiety could impair the recognition of happy facial expressions. As noted earlier, recognizing facial expressions requires applying conceptual knowledge about others and emotions (Adolphs, 2002). Based on their body of research, Leppänen and Hietanen (2004) contended that the advantage for happy faces resulted from "a tendency to form positively biased hypotheses about reality and, especially, about other people" (pp. 27-28). If positive information about other people and their emotions is less chronically accessible or less easily retrieved, then it will take longer to apply emotional knowledge to the perceptual representation of the face. This will manifest in the speed of recognizing an emotion. People high in social anxiety view other people as critical and threatening (Rapee & Heimberg, 1997; Turk et al., 2001). This bias should slow down the recognition of positive facial expressions, given that signals of acceptance and reward are inconsistent with the expected emotions and intentions of others. Thus, the different socioemotional knowledge held by people high and low in social anxiety will translate into differences in the recognition of emotional expressions. THE PRESENT EXPERIMENTS The present experiments examined whether social anxiety biases the recognition of emotional expressions. In each experiment, people completed a two-choice recognition task in which they had to classify emotional expressions as quickly as possible. To maximize the ecological validity of the findings, we used images of real faces displaying prototypical facial expressions (Ekman & Friesen, 1976) instead of iconic or schematic emotional faces. The first experiment compared happy and sad faces; the second experiment compared happy and angry faces. In each experiment, we expected social anxiety to reduce the happy-face recognition advantage. It was unclear at the outset if this would manifest as a smaller advantage for happy faces, a lack of an advantage, or even an advantage for negative emotional expressions. EXPERIMENT 1 Experiment 1 compared the recognition of happy faces and sad faces. Past research with general-population samples has found a happy-face advantage relative to sad faces (Leppänen & Hietanen, 2004; Leppanen, Milders, Bell, Terriere, & Hietanen, 2004). Relative to angry faces, sad faces have not received as much attention in research on general anxiety and social anxiety. Recent research finds that sad faces have many of the same effects as angry faces (see Silvia & Warburton, 2006). Sad faces are detected more efficiently in visual-search paradigms (Eastwood, Smilek, & Merikle, 2001, 2003), and they seem to influence the control of attention (Fenske & Eastwood, 2003). Thus, it seems worthwhile to contrast the recognition of happy and sad faces.