تشخیص حالت چهره در اختلال وسواسی-اجباری

Abstract Sprengelmeyer et al. [Sprengelmeyer, R., Young, A. W., Pundt, I., Sprengelmeyer, A., Calder, A. J., Berrios, G., et al. (1997). Disgust implicated in obsessive–compulsive disorder. Proceedings of the Royal Society of London, 264, 1767–1773] found that patients with OCD showed severely impaired recognition of facial expressions of disgust. This result has potential to provide a unique window into the psychopathology of OCD, but several published attempts to replicate this finding have failed. The current study compared OCD patients to normal controls and panic disorder patients on ability to recognize facial expressions of negative emotions. Overall, the OCD patients were impaired in their ability to recognize disgust expressions, but only 33% of patients showed this deficit. These deficits were related to OCD symptom severity and general functioning, factors that may account for the inconsistent findings observed in different laboratories.

1. Introduction Accurate recognition of facial expressions is a critical element of humans’ social structure (Ekman, 1992), serving as a guide for social behaviour. Even toddlers gain important information from the facial expressions of others (La Barbera, Izard, Vietze, & Parisi, 1976; Sorce, Emde, Campos, & Klinnert, 1985; Young-Browne, Rosenfeld, & Horowitz, 1977). Adults across various cultures recognize six basic facial expressions of emotion: anger, disgust, fear, happiness, sadness, and surprise (Ekman, Levenson, & Friesen, 1983; Ekman, Sorenson, & Friesen, 1969; Izard, 1971), a finding that is reliable across numerous procedural variations (Boucher & Carlson, 1980; Izard, 1971). Despite this apparent universal ability, individuals with some forms of psychopathology are impaired in their recognition of facial expressions, although the specific quality and meaning of these deficits is still poorly understood. Researchers have documented abnormalities in facial expression recognition in alcoholism (Kornreich et al., 2001a and Kornreich et al., 2001b), Alzheimer's disease (Hargrave, Maddock, & Stone, 2002), anorexia nervosa (Zonnevijlle-Bendek, van Goozen, Cohen-Kettenis, van Elburg, & van Engeland, 2002), bipolar disorder (Ketter & Lembke, 2002), criminal psychopathy (Kosson, Suchy, Mayer, & Libby, 2002), social phobia (Simonian, Beidel, Turner, Berkes, & Long, 2001), major depression (Rubinow & Post, 1992), schizophrenia (Addington & Addington, 1998; Archer, Hay, & Young, 1992; Heimberg, Gur, Erwin, Shtasel, & Gur, 1992), and obsessive–compulsive disorder (Sprengelmeyer et al., 1997), the subject of this investigation. Even in the case of depression and schizophrenia, which have been more extensively studied, the precise nature of the deficits is unclear. Some studies report generally impaired recognition of facial expressions among those with major depression (Persad & Polivy, 1993) or schizophrenia (Lewis & Garver, 1995), which would suggest that the impairments reflect a general deficit in face processing or overall neurocognitive functioning, rather than expression recognition per se (Addington & Addington, 1998; Bryson, Bell, & Lysaker, 1997; Kerr & Neale, 1993). More typically, however, results point to problematic recognition of specific facial expressions in depression ( Mikhailova, Vladimirova, Iznak, Tsusulkovskaya, & Sushko, 1996; Rubinow & Post, 1992) and schizophrenia ( Davis & Gibson, 2000; Dougherty, Bartlett, & Izard, 1974; Kucharska-Pietura & Klimkowski, 2002; Muzekari & Bates, 1977; Walker, Marwit, & Emory, 1980). Notably, in a study with normal undergraduates, Rozin, Taylor, Ross, Bennette, and Hejmadi (2005) observed wide variability across participants in general ability to classify emotions depicted in facial expression but no evidence of individual differences in specific recognition deficits. Although affect recognition deficits have been observed in both schizophrenia and depression, the findings show important differences with potential clinical implications. The impairments in the recognition of facial expressions are more severe in schizophrenia than in depression (Feinberg, Rifkin, Schaffer, & Walker, 1986; Heimberg et al., 1992). Impairment in affect recognition appears to improve upon symptom remission in depression (Mikhailova et al., 1996) but not in schizophrenia (Addington & Addington, 1998; Gaebel & Woelwer, 1992), although the deficits predict poor treatment outcome in depression (Geerts & Bouhuys, 1998). In the case of obsessive–compulsive disorder (OCD), researchers have speculated that disgust may play a role in some forms of the disorder, particularly contamination-based types (Phillips, Senior, Fahy, & David, 1998a; Power & Dalgleish, 1997; Woody & Teachman, 2000). In 1997, Sprengelmeyer et al. tested 12 participants with OCD (primarily checking symptoms), 12 with Tourette's Syndrome (five of whom also showed prominent obsessive–compulsive behaviours), 8 anxiety disorder controls, and 40 normal controls on two tasks of facial expression recognition. OCD patients showed a marked deficit in recognition of disgust expressions and normal performance on expressions of anger, fear, happiness, sadness, and surprise. Intriguingly, Tourette's patients with prominent obsessive–compulsive behaviours demonstrated the disgust recognition impairment, while those without obsessive–compulsive behaviours did not. Sprengelmeyer et al. included a control task to rule out reluctance to choose the label “disgust” on the part of individuals with OCD. The findings presented by Sprengelmeyer et al. (1997) have attracted attention not only for their pattern of results but also for their magnitude. Every individual with OCD was impaired in the recognition of disgust, whereas no participant without clinically significant obsessive–compulsive behaviours showed this impairment. Moreover, OCD patients showed specific impairment in the recognition of disgust, rather than a general pattern of poor performance on the task. Some observers have suggested that the effect may occur only among a subset of individuals with OCD—perhaps those with contamination concerns ( Power & Dalgleish, 1997; Woody & Tolin, 2002). Other researchers have pointed to brain functioning in OCD, with imaging results thus far pointing toward the basal ganglia and anterior insula ( Phillips, Young et al., 1998; Phillips et al., 1997; Sprengelmeyer, Rausch, Eysel, & Przuntek, 1998). Despite the unusually strong effect observed by Sprengelmeyer and his colleagues, no other research team has replicated the result. Parker, McNally, Nakayama, and Wilhelm (2004) used procedures that were very close to those used by Sprengelmeyer, with the addition of new models for the facial expressions. They found no overall differences in performance between the OCD and normal control groups. Buhlmann, McNally, Etcoff, Tuschen-Caffier, and Wilhelm (2004) reported similar performance for individuals with OCD and normal controls on a recognition task of prototypical emotional expressions. Although Rozin et al. (2005) used a normal sample, they conducted an analysis examining 26 individuals who scored in the clinical range on the Obsessive Compulsive Inventory (Foa, Kozak, Salkovskis, Coles, & Amir, 1998). These individuals actually recognized disgust expressions better than did participants with lower scores. The Sprengelmeyer et al. (1997) study was intriguing, but without replication it simply remains mysterious. Several of the replication attempts described above were underpowered by ordinary standards, although the initial Sprengelmeyer results yielded such large effects that even very small samples would be expected to replicate them. Parker et al. (2004) raised one interesting clue. Although they failed to replicate the overall results, Parker et al. described one individual, the most severe OCD patient in the sample, who showed marked impairment in recognition of disgust, suggesting that severity of OCD may be an important element of the effect. Unfortunately, severity of symptoms was not detailed in the Sprengelmeyer et al. (1997) study. The study also provided little information on other diagnostic issues such as how diagnosis was determined, which comorbid conditions were present, or types of OCD symptoms beyond checking. The present study aimed to replicate the Sprengelmeyer et al. finding, using a sample of individuals with OCD who were seeking treatment in anxiety specialty clinics as well as adding methodological controls such as structured clinical interviews for diagnosis and symptom severity assessment. Because individuals with depression show deficits in the recognition of facial expressions of emotion, we also examined the role of comorbid depression. Although we adhered closely to the method presented by Sprengelmeyer et al. in many ways, we changed the stimuli to test the robustness of the finding. This study used two models (one male and one female) from the Ekman and Friesen (1976) set, neither of which was the one used by Sprengelmeyer et al. The most common error for Sprengelmeyer's OCD participants was to select anger in place of disgust when naming facial expressions, an error also observed in normal samples (Ekman & Friesen, 1976; Rozin et al., 2005). The Sprengelmeyer et al. stimuli were prepared by morphing each facial expression with two others to increase the difficulty of the task. In the case of disgust stimuli, the expression was morphed with anger or sadness, but not fear. Stimuli in the current study included all possible morphed combinations of disgust, anger, fear and sadness although only the prototypical facial expressions were analyzed

3. Results The primary question of interest in this study was whether OCD patients differ from those with panic and from normal controls in their ability to recognize facial expressions of disgust. Frequency of correct facial expression identifications was tabulated for each participant, across trials for each of the four pure facial expressions. Hit rate was the dependent variable in 3 × 4 repeated measures ANOVA, with one within-subjects factor (stimulus expression: anger, disgust, fear, and sadness) and one between-groups factor (diagnostic group: OCD, panic, and normal control).1 This analysis was accompanied by planned nonorthogonal contrasts, described below. If OCD patients are specifically impaired in their recognition of the expression of disgust, the analysis should show an interaction of facial expression with diagnostic group. The Greenhouse-Geisser adjustment corrected for a violation of the assumption of sphericity. In addition, due to the violation of sphericity, four participants from the OCD group were randomly dropped from the analyses of variance to form equal size groups (Tabachnick & Fidell, 2001).2 3.1. Analysis of variance Table 1 presents the frequency of correct identification of the prototypical facial expressions for each diagnostic group; the maximum number of correct responses was 24. Results of the ANOVA revealed a significant main effect of stimulus expression, F(2, 226) = 19.16, p < 0.0001, View the MathML sourceηp2=0.15, which was qualified by a significant interaction, F(2, 226) = 2.70, p < 0.05, View the MathML sourceηp2=0.05. Simple main effects analyses revealed a significant effect of stimulus expression for the OCD group, F(3, 226) = 11.32, p < 0.0001, View the MathML sourceηp2=0.13, and for the panic group, F(3, 226) = 3.70, p < 0.05, View the MathML sourceηp2=0.05. Tukey's method of post hoc comparisons revealed that the OCD group was significantly less accurate at the recognition of disgust relative to all other expressions (fear: q = 8.10, p < 0.001; anger: q = 5.33, p < 0.01; sadness: q = 4.69, p < 0.01). In addition, Tukey's method revealed that panic patients recognized fear significantly more accurately than they recognized disgust (q = 4.16, p < 0.05). Their accuracy at recognizing disgust was not different from their recognition of sadness (q = 2.72, p > 0.10) or anger (q = 0.59, p > 0.10). There was no main effect for diagnostic group, F(2, 105) = 1.65, p > 0.10, View the MathML sourceηp2=0.03. Table 1. Mean (and standard deviation) frequency of correct identification of facial expressions of emotions by diagnostic group Expression Participant group OCD Panic Normal Anger 22.14 (2.67) 21.69 (2.36) 21.61 (2.97) Disgust 19.36 (4.89) 21.39 (4.03) 21.86 (3.21) Fear 23.58 (1.00) 23.56 (1.13) 23.50 (1.16) Sadness 21.81 (3.40) 22.81 (1.77) 22.36 (2.64) Note: These stimuli were taken directly from the Ekman and Friesen set of emotional expressions; participants rated 24 of these photographs. Table options 3.2. Planned comparisons Planned nonorthogonal contrasts were conducted in order to test specific hypotheses not tested directly by the omnibus ANOVAs. Whereas the overall F-test yields information about whether there are any differences among the groups, planned contrasts permit specific predictions to be tested with relatively strong statistical power. The number of contrasts was limited to two in order to maintain control over Type I error ( Rosenthal & Rosnow, 1985). The first contrast compared the three groups specifically on disgust recognition. The OCD group performed this task significantly less accurately than the other two groups, F(1, 105) = 10.78, p < 0.01, r = 0.31. The second contrast compared recognition of disgust facial expressions in the OCD group against all other emotional expressions and other groups. The OCD group recognized disgust expressions less accurately than all other groups and all other emotional expressions, F(1, 331) = 29.45, p < 0.00001, r = 0.28. 3.3. Predictors of poor performance Fig. 1 shows the distribution of hit rates for recognition of pure disgust expressions across the three groups. Most individuals in the panic and normal control groups showed excellent ability to recognize disgust, with the mean coming close to 100% accuracy despite the presence of several individuals who performed quite poorly. The distribution for the OCD group, however, was much more variable. In an effort to examine potential predictors of poor performance, OCD participants were divided into two groups: those who recognized pure (prototypical) facial expressions of disgust within normal limits and those who were clearly impaired at recognizing disgust. Performing within normal limits was arbitrarily defined as demonstrating a hit rate within 1.5 standard deviations of the mean of the normal comparison group. Twenty seven individuals fell into this group, with a mean hit rate of 22.56 (S.D. = 1.69). 13 individuals were clearly impaired, falling more than 1.5 standard deviations below the mean of the normal comparison group in the recognition of prototypical facial expressions of disgust (hit rate M = 12.92, S.D. = 3.43). Hit rate for recognition of disgust facial expressions across diagnostic groups. Fig. 1. Hit rate for recognition of disgust facial expressions across diagnostic groups. Figure options Using this definition for impaired recognition performance described above, in addition to the 13 participants with impaired performance in the OCD group, three participants in the normal comparison group and four participants in the panic group demonstrated impaired functioning on this task. Their observations can be seen as within-group outliers in Fig. 1. This difference in distribution of impaired performance was significant, χ2(2, N = 112) = 9.19, p < 0.01. Similar numbers of participants (<5) in each group showed impaired recognition of facial expressions other than disgust, but the distribution of these deficits was equivalent across groups. Using a series of t-tests, OCD patients who performed normally on the pure expression task were compared to those who showed a deficit in an effort to discover descriptive or illness-related factors that correlate with impaired recognition of disgust expressions. Sample sizes differ for each t-test due to missing data. Severity of symptoms was of particular interest, given the results reported by Parker et al. (2004). Symptom severity indicators used in these analyses were: (a) Global Assessment of Functioning (GAF; the Axis V indicator of general functioning and illness severity gleaned from the diagnostic interview and rated by the interviewer), (b) YBOCS Total Score, (c) YBOCS Compulsions subscale score, (d) YBOCS Obsessions subscale score, (e) BDI score, and (f) duration of OCD in years. Participants’ current age was also explored as a possible predictor, as age has been found to be related to performance on facial expression judgment tasks ( McDowell, Harrison, & Demaree, 1994). One-tailed tests were used for t-tests due to the a priori hypotheses that performance would be negatively affected by greater severity of symptoms and higher age. Table 2 shows the results of this analysis. OCD patients with normal disgust recognition were significantly more functional and less symptomatic than those who were impaired in their recognition of disgust facial expressions. OCD patients who were impaired on the disgust recognition task had significantly higher scores on the YBOCS (both subscales as well as the total score) and significantly lower GAF ratings. Results of the t-tests on age of onset, duration of OCD, current age, and disgust sensitivity did not reveal significant differences. Table 2. Comparison of individuals with OCD who showed normal versus impaired recognition of disgust facial expressions Variable Normal mean (S.D.) Impaired mean (S.D.) t d.f. r GAF 57.08 (7.66) 53.00 (8.54) 1.47† 36 0.21 YBOCS total 21.96 (4.44) 25.08 (4.29) −2.02* 35 −0.30 YBOCS obsessions 11.30 (2.27) 12.73 (2.00) −1.78* 32 −0.25 YBOCS compulsions 10.74 (3.11) 13.00 (2.57) −2.09* 32 −0.34 BDI 15.6 (8.97) 18.78 (13.87) −0.74 27 −0.19 Duration of OCD 15.3 (11.33) 17.36 (12.56) −0.48 32 0.00 Age 34.12 (11.71) 36.62 (13.82) −0.59 37 −0.02 Disgust sensitivity 16.51 (6.47) 15.79 (5.96) 0.27 24 −0.12 Note: †p < 0.10, *p < 0.05 (one-tailed); GAF: Global Assessment of Functioning (SCID-I); YBOCS: Yale-Brown Obsessive Compulsive Scale; BDI: Beck Depression Inventory. Correlations refer to hit rates; all are in the expected direction. Table options 3.4. Role of depression Examining the hit rates for pure disgust expressions, which showed a larger effect size than the morphed expressions, comorbid depression does not account for the poor recognition rates exhibited by the OCD group. The correlation between recognition rate of pure disgust expressions and the BDI was r = −0.19. Individuals who were impaired in their recognition of disgust were not more likely to be diagnosed with current or partially remitted depression, χ2(1, N = 40) = 1.05, p > 0.30. 3.5. Role of symptom profile Individuals with OCD were classified according to their primary symptom in an effort to discover whether contamination concerns (which some researchers hypothesize would be more related to disgust perception) or checking rituals (which was the primary symptom experienced by individuals in the Sprengelmeyer et al. study) would be related to impaired recognition of disgust expressions. Among participants with primary contamination concerns, 22% showed impaired performance on the task. In comparison, 18% of participants with primary checking rituals showed impaired performance. A larger percentage of individuals with primary obsessions (45%) showed impaired performance, but with such a small sample, this difference was not statistically significant, χ2(2, N = 31) = 2.27, p > 0.30. Other types of OCD symptoms were not analyzed because of low frequency. Participants in the three primary symptom groups did not differ with regard to severity of OCD symptoms, F(2, 29) = 1.38, or global functioning, F(2, 30) = 1.77.