بازشناسی حالت چهره عاطفی دچار اختلال در افراد مبتلا به اعتیاد به الکل در مقایسه با اختلال وسواسی-اجباری و کنترل های طبیعی

Abstract Emotional facial expression (EFE) decoding skills have been shown to be impaired in recovering alcoholics (RA). The aim of the present study is to replicate these results and to explore whether these abnormalities are specific to alcoholism using two control groups: non-patient controls (NC) and patients with obsessive-compulsive disorder (OC). Twenty-two alcoholic patients at the end of their detoxification process (RA) were compared to 22 OC and 22 NC matched for age, sex and education level. They were presented with 12 photographs of facial expressions portraying different emotions: happiness; anger; and fear. Each emotion was displayed with mild (30%) and moderate (70%) intensity levels. Each EFE was judged on 8 scales labeled happiness, sadness, fear, anger, disgust, surprise, shame and contempt. For each scale, subjects rated the estimated intensity level. RA were less accurate in EFE decoding than OC and NC, particularly for anger and happiness expressions. RA overestimated the emotional intensity for mild intensity level expressions compared with both OC and NC while no significant differences emerged for moderate intensity level expressions. Deficits in EFE decoding skills seem to be specific to RA when compared with OC. Comparison with other psychopathological groups is still needed. Possible consequences of EFE decoding deficits in RA include distorted interpersonal relationships.

. Introduction Emotional facial expression (EFE) recognition has been shown to be severely impaired in recovering alcoholics (RA) (Philippot et al., 1999). RA made significantly more errors in identifying the emotion displayed by a face than non-patient controls, with a special bias toward overattribution of anger and contempt. RA also systematically overestimated the intensity of the emotions portrayed by the faces. Moreover, RA did not perceive this decoding deficit. These impairments seemed to remit only partially with abstinence (Kornreich et al., in press), as decoding biases for anger and disgust were present in mid- to long-term abstinent patients while intensity overestimation disappeared. Clinical implications of EFE recognition deficit could involve interpersonal relationship impairments since the decoding of non-verbal cues constitutes an essential process in normal communication and interaction regulation (Patterson, 1999). The aim of the present study is to confirm our previous findings in RA and to explore if the observed decoding EFE pattern is specific to this population. Indeed, a major methodological consideration in alcoholism research involves the ability to show that an observed effect is specific to alcoholism. To demonstrate such specificity, alcoholics must differ on the studied dimension from control groups with other behavior problems as well as from ‘non-patient’ control groups (Sher et al., 1999). At this stage, two non-alcoholic control groups were therefore included in the present study, one with psychopathology, namely obsessive-compulsive disorder (OCD), and one with no psychopathology. We chose an OCD control group because alcoholism and OCD display symptomatic similarities but do not share common etiologies. In particular, several investigators have noted similarities between urges and desires to abuse alcohol and OCD. Researchers in the field of alcoholism have characterized alcohol abusers as having a ‘compulsion’ to use alcohol (Edwards and Gross, 1976, Caetano, 1985, Modell et al., 1992 and Roberts et al., 1999). It has also been suggested that the craving for alcohol seen in alcohol abusers resembles obsessive thought patterns (Modell et al., 1992 and Anton et al., 1995). However, the life-time risk for OCD among close relatives of alcoholics is 1.4%, which does not support the existence of a common genotype for the two disorders (Schuckit et al., 1995). Therefore, OCD appears relevant as a psychopathological control group.

3. Results Statistical analyses did not reveal any significant correlation between depression or anxiety scores, on the one hand, and decoding accuracy or intensity scores, on the other hand. Therefore, it was not necessary to control for their effects (e.g. with analysis of covariance). Likewise, no main significant gender effect or interaction was observed. Therefore, all subsequent analyses were collapsed across these factors. Finally, for RA, and for OC, no significant correlation emerged between personal history of alcoholism (i.e. number of drinks per day, duration of illness, number of previous in-patient detoxification treatments and positive familial antecedents for alcoholism) (Table 1) or personal history of OCD (i.e. age of onset, duration of illness, number of previous treatments and positive familial antecedents for OCD) (Table 1) and decoding accuracy or intensity scores. 3.1. Decoding accuracy In order to assess whether RA showed a difference in the ability to decode EFE compared with NC (non-patient controls) and OC (OCD controls), a repeated measures analysis of variance using a multivariate approach was conducted on the accuracy scores with emotion (happiness, anger and fear) and intensity level (30%, 70%) as within-subject factors and group (RA, NC and OC) as between-subjects factors. The results are shown in Table 2. Table 2. F-values for MANOVA, (accuracy scores) Sources d.f. F-values η2 Power Group 2,63 6.18a 0.164 0.877 Emotion 2,62 51.01b 0.622 1.000 Intensity level 1,63 55.77b 0.470 1.000 Emotion×group 4,126 2.07c 0.062 0.604 Intensity level×group 2,63 5.81a 0.156 0.855 Intensity level×emotion 2,126 1.58 0.210 0.329 Intensity level×emotion×group 4,126 0.14 0.004 0.079 Emotion (happiness, anger and fear) and intensity level (30%, 70%) as within-subject factors, and group (RA, NC and OC) as between-subjects factors. a P<01. b P<0.001. c Marginally significant interaction: P=0.088. Table options Overall, RA (m=0.147, S.D.=0.17) are significantly less accurate than NC (m=0.318, S.D.=0.20) and OC (m=0.322, S.D.=0.17). No significant differences emerged between OC and NC. Post-hoc analyses were conducted to investigate the group×intensity level interaction. For 70% expression, RA were significantly less accurate than NC, F(1,42)=10.08, P=0.003 (m=0.181, S.D.=0.23 and m=0.431, S.D.=0.28, respectively) while the difference was marginally more significant for 30% expression, F(1,42)=3.55, P=0.066 (m=0.113, S.D.=0.15 and m=0.204, S.D.=0.16, respectively). When RA were compared with OC, for 70% expression RA were significantly less accurate than OC, F(1,42)=14.35, P<0.001 (m=0.181, S.D.=0.23 and m=0.454, S.D.=0.28, respectively), while the difference was not significant for 30% expression. Means for these analyses are shown in Fig. 2. Decoding accuracy scores as a function of group, facial expressions and ... Fig. 2. Decoding accuracy scores as a function of group, facial expressions and intensity level (mild=30% and moderate=70% expressions) showing significant differences in accuracy for happiness and anger expressions between RA (recovering alcoholics) on the one hand and OC (obsessive-compulsive disorder patients) and NC (nonpatient control subjects) on the other hand. Figure options Post-hoc analyses were conducted to investigate the group×emotion interaction. No significant differences emerged between OC and NC. Regarding differences between RA and NC, previous research (Philippot et al., 1999 and Kornreich et al., in press) has shown differences on accuracy scores as a function of the emotion portrayed by the faces. Specifically, RA were less accurate for the decoding of happiness, anger, sadness and disgust while no difference was found for fear. In order to test these results, post-hoc analyses were conducted. The results revealed that RA were significantly less accurate than NC in the decoding of the expression of happiness F(1,42)=7.57, P=0.009 (m=0.31, S.D.=0.37 and m=0.62, S.D.=0.36, respectively) and anger F(1,42)=4.61, P=0.038 (m=0.06, S.D.=0.11 and m =0.21; S.D.=0.30, respectively) while no difference emerged for fear. Thus, previous findings were replicated. 3.2. Intensity scores Repeated measure analysis of variance using a multivariate approach was conducted on the intensity scores with emotion (happiness, anger and fear), intensity level (30% and 70%) and scales (happiness, sadness, fear, anger, disgust, surprise, shame and contempt) as the within-subject factors, and group (RA, NC and OC) as the between-subjects factor. Table 3 displays the results for all significant main effects and interactions. However, in the context of the present article, only main effects or interactions involving group are of interest and we shall limit our presentation and discussion to these results. As can be seen in Table 3, among the interactions involving group, the group×intensity level interaction accounted for the largest percentage of variance (as indicated by η2). Therefore, post-hoc analyses focused on the group×intensity level interaction and not on the subsequent second and third order interactions, which explain less variance. Table 3. F-values for MANOVA (intensity scores) Sources d.f. F-values Power η2 Group 2,63 2.85 0.539 0.083 Emotion 2,126 16.88c 1.00 0.211 Intensity level 1,63 10.16b 0.882 0.139 Scales 7,441 8.97c 1.00 0.125 Group×intensity level 2,63 4.07a 0.704 0.115 Emotion×intensity level 2,126 11.14c 0.991 0.150 Emotion×group 4,126 0.68 0.217 0.021 Intensity level×scales 7,441 22.65c 1.00 0.265 Emotion×scales 14,882 30.93c 1.000 0.329 Group×scales 14,441 1.16 0.719 0.035 Group×emotion×scales 28,882 1.95b 0.998 0.059 Group×intensity level×emotion 4,126 2.14 0.620 0.064 Group×intensity level×scales 14,441 0.73 0.467 0.022 Emotion×intensity level×scales 14,882 24.00c 1.00 0.276 Group×emotion×intensity level×scales 28,882 1.85b 0.996 0.055 Emotion (happiness, anger and fear), intensity level (30% and 70%) and scales (happiness, sadness, fear, anger, disgust, surprise, shame and contempt) as within-subject factors, and group (RA, NC and OC) as between-subject factors. a P<0.05. b P<0.01. c P<0.001. Table options Post-hoc analyses were conducted to investigate the group×intensity level interaction. No significant differences emerged between OC and NC. When RA were compared with NC, the results revealed that for 30% expression RA overestimated the overall emotional intensity, F(1,42)=4.71, P=0.036 (m=2.82, S.D.=0.83 and m=2.34, S.D.=0.60, respectively) while the difference was not significant for 70% expression (m=2.77, S.D.=0.56 and m=2.61, S.D.=0.70, respectively). Similarly, when RA were compared with OC, the results revealed that for 30% expression RA overestimated the overall emotional intensity, F(1,42)=7.03, P=0.011 (m=2.82, S.D.=0.83 and m=2.19, S.D.=0.73, respectively) while the difference was not significant for 70% expression (m=2.77, S.D.=0.56 and m=2.50, S.D.=0.61, respectively). Means for these analyses are shown in Fig. 3 and overall intensity ratings are displayed in Table 4. Mean emotional intensity evaluation scores as a function of group and intensity ... Fig. 3. Mean emotional intensity evaluation scores as a function of group and intensity levels (mild or moderate). Figure options Table 4. Means for intensity scores in function of the EFE, intensity level and groups Intensity level (%) Happiness Sadness Contempt Shame Anger Disgust Surprise Fear Alcoholics Anger 30 1.93 (0.88) 3.61 (1.70) 3.23 (1.79) 2.75 (1.56) 2.57 (1.18) 3.05 (1.68) 2.70 (1.62) 2.52 (1.40) Happiness 30 2.82 (1.76) 3.07 (1.68) 2.68 (1.67) 2.29 (1.35) 3.11 (1.89) 2.57 (1.45) 2.39 (1.42) 2.43 (1.52) Fear 30 2.66 (1.69) 4.61 (1.51) 2.77 (1.57) 3.20 (1.82) 2.41 (1.48) 2.91 (1.71) 2.36 (1.43) 3.14 (1.31) Anger 70 2.18 (1.12) 3.14 (1.26) 3.57 (1.92) 2.02 (1.08) 3.52 (1.92) 3.20 (1.70) 2.11 (1.39) 2.11 (1.17) Happiness 70 3.59 (2.63) 1.93 (1.36) 2.50 (1.89) 1.82 (1.47) 3.73 (2.63) 2.41 (1.69) 2.41 (1.39) 2.23 (1.73) Fear 70 1.68 (1.07) 3.57 (1.99) 3.25 (1.90) 2.39 (1.71) 2.36 (1.42) 3.27 (1.93) 3.86 (1.91) 3.82 (2.09) OCD Anger 30 1.80 (0.87) 3.07 (1.58) 2.36 (1.61) 1.91 (1.11) 2.18 (1.34) 2.11 (1.46) 1.95 (1.33) 2.43 (1.34) Happiness 30 3.20 (1.39) 2.45 (1.24) 2.09 (1.24) 1.70 (0.87) 2.00 (1.46) 1.93 (1.19) 1.61 (0.71) 1.89 (1.09) Fear 30 1.93 (0.78) 3.23 (1.39) 2.20 (1.39) 2.07 (0.93) 1.93 (0.78) 2.09 (1.27) 1.95 (1.08) 2.54 (1.32) Anger 70 1.07 (0.18) 2.68 (1.32) 4.00 (1.73) 2.02 (1.38) 4.66 (1.30) 3.18 (1.66) 2.05 (1.45) 2.27 (1.17) Happiness 70 4.82 (2.20) 1.32 (0.48) 1.82 (1.34) 1.29 (0.50) 1.98 (2.08) 1.34 (0.82) 1.61 (0.79) 1.48 (1.13) Fear 70 1.36 (0.73) 2.61 (1.52) 2.11 (1.26) 2.16 (1.28) 2.48 (1.59) 2.70 (1.55) 4.50 (1.85) 4.66 (1.75) Normal controls Anger 30 1.89 (0.74) 3.07 (1.11) 2.68 (1.28) 1.93 (.93) 2.80 (1.20) 2.43 (1.14) 1.86 (0.99) 2.36 (1.09) Happiness 30 3.57 (1.74) 2.39 (1.24) 2.34 (1.21) 1.75 (1.13) 2.18 (1.63) 1.86 (1.16) 1.77 (1.09) 2.02 (1.24) Fear 30 2.43 (1.13) 3.18 (1.32) 2.68 (1.48) 2.04 (1.08) 1.98 (1.14) 2.34 (1.42) 2.14 (1.16) 2.59 (1.34) Anger 70 1.34 (0.88) 2.45 (1.57) 3.70 (1.97) 1.91 (1.39) 4.27 (2.02) 3.11 (1.57) 2.11 (1.19) 2.11 (1.19) Happiness 70 4.98 (2.25) 1.43 (0.85) 1.68 (1.05) 1.43 (0.78) 2.34 (2.20) 1.36 (0.64) 2.09 (1.13) 1.59 (0.99) Fear 70 1.39 (0.75) 3.11 (1.74) 3.04 (1.89) 2.11 (1.40) 2.61 (1.68) 3.00 (1.79) 4.36 (1.93) 4.84 (2.03) Note 1: each of the 12 expressions was rated by the participants on seven-point intensity scales (from 1: not at all; to 7: extremely). Table options 3.3. Difficulty ratings Repeated measure analysis of variance using a multivariate approach was conducted on difficulty ratings with emotion (happy, anger and fear), and intensity level (30% and 70%) as the within-subjects factors, and group (RA, NC and OC) as the between-subjects factor. The results revealed neither significant main effects nor interactions involving group, meaning that RA were not aware of their deficits.