This study confirmed the hypothesis that individuals with BDD are slower and less accurate than controls at identity-matching of faces with emotional expressions. The BDD group had more than twice the error rate for the matching task with emotional faces compared to the healthy controls. The BDD group showed the greatest difference in reaction time from healthy controls for emotional faces, followed by neutral faces and then ovals/circles. However, there was no differential effect on the BDD group of any specific emotion type. In total, these data suggest that individuals with BDD have abnormalities in the speed and accuracy of processing faces with emotional expressions. This builds on findings from previous studies of abnormal interpretation of emotions (Buhlmann et al. 2004
; Buhlmann et al. 2006
), to suggest that there may be more fundamental abnormalities for perception of faces with emotional expressions.
These results may be explained in that the matching condition for emotional faces differed from the matching condition for neutral faces on several dimensions that may have affected performance for the BDD subjects. Although the faces had emotional expressions, the explicit task was not to interpret the expression but to match the face with the neutral face of the same person. For the neutral face task, correct choices were identical. The emotional face task, however, required processing of facial features to match the emotional face to the neutral face. This identification process normally involves visual analysis of both details and configural aspects of faces whereas matching neutral faces could be accomplished either by matching configural information (Vuilleumier et al. 2003
) or by matching one or more details. Previous neuropsychological testing demonstrated that on a visuospatial task individuals with BDD they tend to focus on isolated details rather than larger, global organizational features (Deckersbach et al. 2000
). In addition, a recent functional magnetic resonance imaging (fMRI) study demonstrated greater left-hemisphere activity in BDD compared to a control group while matching identities of others’ neutral-expression faces (Feusner et al. 2007
). Although further research is needed to verify this, one possibility is that this finding in BDD may be associated with an imbalance in detailed vs. holistic processing. If BDD subjects overly rely on details for processing emotional faces as well, this slower strategy (Peyrin et al. 2006
) may account for delayed reaction times relative to controls. It may also prove to be less accurate, at least within the limited stimulus presentation timeframe in this experiment of 4 seconds. These possible explanations remain to be tested directly. However, the fact that we did not find a significant group by stimulus effect for the different types of emotion supports a face-processing deficit that occurs for faces with emotional expressions in general, rather than an influence of emotion per se, for which we would expect a differential influence on the BDD group depending on the valence of emotion.
Concerning facial processing in general, less is known in BDD compared to social phobia. However, the two disorders share many clinical features including fears of negative evaluation, rejection, ridicule by others (Wilhelm et al. 1997
; Phillips et al. 1998
), and ideas of reference (American Psychiatric Association. 2000
). Several studies of social phobia have demonstrated abnormalities in the processing of emotional faces (Simonian et al. 2001
; Horley et al. 2004
; Juth et al. 2005
). In a study of children with social phobia, Simonian et al., found impaired explicit recognition of facial affect that included happy, disgusted, and angry faces (Simonian et al. 2001
). Juth, et al. (2005)
studied the relationship of facial emotional expression and social anxiety in both healthy and social phobia subjects (Juth et al. 2005
). In individuals with or without social anxiety, the processing of faces with fearful or angry expressions was slower and less accurate than for happy faces. In addition, individuals with a diagnosis of social phobia were less accurate than healthy controls in facial identification for fearful and happy averted target faces. Although the explicit task was different, we found similar results of lower accuracy of facial identification in the BDD group for emotional faces. One difference in our study, the significance of which is unclear, is that mean RT for angry faces was not significantly different than for happy faces.
There are other possibilities as to why the BDD group had slower responses to faces with emotional expressions during the matching task. The mean RT for the BDD group relative to the controls progressively diverged from non-face objects (ovals and circles) to neutral faces, to emotional faces (). This could indicate that the degree of general salience of the stimuli corresponded with progressively slower RT. This occurred for both groups, although it was more pronounced for the BDD group. Several studies have shown that the processing of threat-related information is highly prioritized and may occur automatically, even when the threat is not explicitly attended (Morris et al. 1998
; Whalen et al. 1998
; Vuilleumier and Schwartz 2001
). In a study of selective processing of threatening information Buhlmann, et al. (2002)
demonstrated that individuals with BDD were more easily distracted than controls by emotional cues in general and most distracted by words related to their disorder, such as “disfigured” or “pretty” (Buhlmann et al. 2002
). Thus, feature processing in the BDD group may also be more susceptible to distraction from the presentation of emotional faces, resulting in slower RTs and more errors. Yet if this were the case, we would expect to see proportionally slower RTs in the BDD group for the more threatening faces (angry, disgusted, or fearful), which we did not observe.
An additional factor that may have influenced processing time for emotional faces is failure in inhibition. To respond quickly and accurately to match emotional faces they must attend to the facial identity while implicitly inhibiting attention to the emotional valence. Maxwell, et al., (2005)
demonstrated in healthy controls that the presentation of task-irrelevant emotional faces (angry and happy) resulted in more inhibitory errors relative to the presentation of neutral faces on an explicit go/no-go task (Maxwell et al. 2005
). Other studies have also shown that the perception of emotional stimuli is able to bias competing information processing schemes (Pessoa et al. 2002
; Bishop et al. 2004
). Similar inhibitory errors from emotional faces in the current study could account for the higher error rates on the matching task for both groups, relative to the neutral faces and ovals/circles. The fact that the BDD group had approximately twice the error rate for the emotional faces suggests that they may have a more marked failure in inhibition than healthy controls.
Another possible explanation for the differences between groups could be related to the complexity of the visual task, and not specifically to face processing. The observation that error rates were greater in the BDD group for emotional faces than neutral faces and non-face objects (ovals and circles) could be due to the fact that the emotional faces task was a more difficult task. Unlike the other two tasks, it did not just involve matching of identical visual constructs. However, Buhlmann et al. (2004)
in a previous study of face processing tested subjects with BDD using the Short Form of the Benton Facial Recognition task (Buhlmann et al. 2004
). This tests facial identity matching using neutral-expression target faces that are the same individuals’ face yet presented at a different angle, and was therefore more difficult than the matching task in the current study. In this study individuals with BDD did not perform significantly differently than controls (mean scores of 23.7±2.5 and 23.3±2.7, respectively). This suggests that the differences seen in the current study may not just be a reflection of the task difficulty. Nevertheless, future studies will still be useful to clarify whether these differences are specific for faces as opposed to other types of complex visual stimuli.
Regardless of the cause, aberrant processing of others’ faces may contribute to the symptomatology in BDD. Abnormal perception of faces with emotional expressions could contribute to misinterpretation of emotions and subsequently lead to ideas of reference in which they believe others are regarding them in a contemptuous or threatening manner. As they are usually concerned about others’ judgments of their appearance, this could result in significant distress. If, in fact, the basis of abnormal emotional face processing is aberrant feature processing and this occurs for their own face as well, this could be an important factor in their apparent perceptual distortions of their own appearance.
One of the limitations of the current study is that we do not know the subjects’ interpretation of each emotional expression, as this was not an explicit part of the task. It is therefore unclear if misinterpretation accounted for differences in performance. Similarly, we did not have subjects rate the aversiveness of faces. Although anxiety levels were not significantly different between groups during the task as a whole, we do not know if there were differences in emotional arousal for specific faces. The fact that our face database only included 2 surprised stimuli, as opposed to 6 for the other emotions, was also a limitation.
Another limitation of this study is the small sample size. This may have limited the power of detection of differences between groups, particularly for the individual emotional expressions. It also limited our ability to analyze the effects of comorbid MDD and GAD diagnoses on RTs and accuracy rates. Future, larger studies that address these variables could provide useful information to further understand the basis of delayed reaction times and higher error rates.
This study’s findings of abnormalities in processing of emotional faces in BDD may reflect fundamental differences in the nature of their visual information processing, or may suggest a specific problem in processing of emotional information conveyed by faces. Thus, the perceptual abnormalities seen in BDD may not be specific to misperception of their own faces but may reflect a more fundamental problem in human face perception. Future studies are needed to elucidate the specific causes of aberrant face processing in BDD.