We investigated the deficit in the recognition of facial emotions in a sample of medicated, stable Korean patients with schizophrenia using Korean facial emotion pictures and examined whether the possible impairments would corroborate previous findings.
Fifty-five patients with schizophrenia and 62 healthy control subjects completed the Facial Affect Identification Test with a new set of 44 colored photographs of Korean faces including the six universal emotions as well as neutral faces.
Korean patients with schizophrenia showed impairments in the recognition of sad, fearful, and angry faces [F(1,114)=6.26, p=0.014; F(1,114)=6.18, p=0.014; F(1,114)=9.28, p=0.003, respectively], but their accuracy was no different from that of controls in the recognition of happy emotions. Higher total and three subscale scores of the Positive and Negative Syndrome Scale (PANSS) correlated with worse performance on both angry and neutral faces. Correct responses on happy stimuli were negatively correlated with negative symptom scores of the PANSS. Patients with schizophrenia also exhibited different patterns of misidentification relative to normal controls.
These findings were consistent with previous studies carried out with different ethnic groups, suggesting cross-cultural similarities in facial recognition impairment in schizophrenia.
Impaired recognition of facial affect in schizophrenia has been documented extensively.
Although schizophrenia appears quite to be similar across a range of cultures, cross-cultural variability has been noted.
Although several studies in Korea have also investigated the ability to recognize facial emotion in schizophrenia, using either non-Korean or less well-validated facial stimuli, these researchers did not fully demonstrate the same-race effect.
The most widely used facial stimuli for evaluating emotion recognition performance were developed by Ekman and Friesen.
In the present study, we investigated deficits in the recognition of facial emotions in a sample of medicated, stable Korean patients with schizophrenia using Korean facial emotion pictures and examined whether the possible impairments would corroborate previous findings. Specifically, neutral faces were included to see if patients with schizophrenia misidentified neutral cues as unpleasant or threatening relative to normal controls.
Fifty-five patients with schizophrenia (24 men, 31 women) were recruited from Uijeongbu St. Mary's Hospital and Kyungpook National University Hospital, and 62 healthy control subjects (29 men, 33 women) were recruited from community settings. After receiving a complete description of the study, written informed consent was obtained from all participants. This study was approved by the Institutional Review Board of Kyungpook National University Hospital.
Patients were clinically stable in- or out-patients without prominent positive symptoms at the time of the study. The diagnosis of schizophrenia was established by experienced psychiatrists according to the DSM-IV. Detailed chart reviews were conducted to rule out patients with mental retardation, with a history of substance use disorder, and with any neurologic and medical disorders known to influencing cognitive functioning.
Regarding antipsychotic medications, 49 patients were taking stable regimens of atypical antipsychotics (amisulpride, aripiprazole, clozapine, olanzapine, quetiapine, or risperidone), and six were taking stable regimens of typical antipsychotics (haloperidol or fluphenazine) at therapeutic dosages. Healthy participants were recruited from staff members in the two hospitals mentioned above. Subjects were screened for psychotic, mood, and substance use disorders as well as for a history of head injury with significant loss of consciousness (greater than 5 min) or neurological disorder.
Demographic data for both groups and symptom characteristics of the schizophrenic patients are shown in
FAIT is a computerized test that uses ChaeLee Korean facial expressions of emotions including happiness, sadness, fear, anger, disgust, surprise, and neutral as stimuli. Methods of its development and validation were reported elsewhere.
In the current study, subjects were given an explanation regarding each emotion and had two practice sessions before the FAIT began. Then, subjects were asked to choose one of six emotions or a neutral expression while viewing the facial images. A total of 44 facial images displaying happy, sad, fearful, angry, surprised, disgusted, or neutral expressions were then presented: seven faces expressed surprise and disgust, and six faces expressed the other emotions and neutral expressions. Hence, the maximum score was either six or seven according to the number of facial images presented for each emotion. The order of stimuli was randomized. The choices were displayed on the screen along with the stimuli, and subjects responded by pressing the touch screen.
The chi-square test (gender) and the t-test (age, education) were used to compare demographic characteristics between the schizophrenia and normal control groups. For the analysis of between-group differences on the FAIT, analysis of covariance (ANCOVA) was used for overall correct responses, with education as a covariate, and multivariate analysis of covariance (MANCOVA) was used for each emotion, with education as a covariate. Analysis of the total mean response time as well as the mean response time to an individual emotion was carried out in the same manner. Pearson's correlations were performed to detect the relationship between psychiatric symptoms and performance on the FAIT in the schizophrenia group. Fisher's exact test was also used to compare error patterns between the two groups.
Data were analyzed using PASW Statistics, version 18 (SPSS, Inc.; Chicago, IL, USA). The significance level was established at 0.05.
The ANCOVA, with education as a covariate, revealed that patients with schizophrenia (29.2±4.9) correctly identified the overall emotion of facial stimuli less often than did controls (32.5±3.7) [F(1,114)=6.33, p=0.013]. The MANCOVA revealed that schizophrenic patients preformed worse than controls in the recognition of sadness, fear, and anger [F(1,114)=6.26, p=0.014; F(1,114)=6.18, p=0.014; F(1,114)=9.28, p=0.003, respectively](
Schizophrenic patients showed significantly longer overall mean response times to facial stimuli than did controls (5.50±2.5 sec for schizophrenic patients; 3.10±1.4 sec for controls) [F(1,114)=17.38, p<0.001]. Relative to healthy participants, schizophrenic patients showed a delayed response to all emotional and neutral faces, except fearful faces (
The correlations between correct response on the FAIT and symptom severity were significant for the total score as well as the three subscale scores of the PANSS. Within emotions, higher total and three subscale scores of the PANSS correlated with worse performance on both angry and neutral faces. The correct responses on the happy stimuli were negatively correlated with negative symptoms on the PANSS. However, only the correlation between angry faces and the PANSS was significant when the Bonferroni correction for multiple comparisons was applied [α-adjusted p level=0.0018 (0.05/28)](
No significant correlations between the mean overall response time on the FAIT and the total as well as three subscale scores of the PANSS were found. However, the correlation between the overall response time and negative symptoms reached significance (r=0.262, p=0.053). Within emotions, higher negative subscale scores on the PANSS were correlated with longer response times for sad, angry, and neutral faces. However, the correlations were not significant when the Bonferroni correction for multiple comparisons was applied [α-adjusted p level=0.0018 (0.05/28)](
Duration of illness was not correlated with total correct response (r=-0.038, p=0.545) or with response time (r=0.136, p=0.323).
We found that schizophrenic patients and healthy subjects differed in the distribution of errors for angry (p=0.014), surprised (p<0.001), disgusted (p=0.012), and neutral expressions (p<0.001; all by Fisher's exact test). In general, whereas error responses in healthy participants were related to one prominent emotion, those in schizophrenic patients were relatively divergent. Schizophrenic patients were less likely to attribute disgust to angry expressions (21% for schizophrenic patients compared with 67% for healthy subjects), fearful to surprised expressions (46% versus 74% respectively) and sadness to disgusted expressions (66% versus 79% respectively) compared with normal controls. Interestingly, schizophrenic patients were more likely to attribute happiness to neutral expressions (28% versus 3% respectively) than were controls. In both groups, sad faces were most commonly misrecognized as disgusted, followed by misrecognizing angry expressions. Fearful faces were most commonly misrecognized as surprised, followed by misrecognizing angry expressions. Angry faces were most commonly misrecognized as disgusted, followed by misrecognizing fearful expressions. Surprised faces were most commonly misrecognized as fearful, followed by misrecognizing angry expressions. Disgusted faces were most commonly misrecognized as sad, followed by misrecognizing fearful expressions. Neutral faces were most commonly misrecognized as sad expressions (
Impaired recognition of facial affect in schizophrenia has been investigated extensively. However, few studies have used color photographs of Korean faces with moderate sample size to investigate these impairments among Korean patients with schizophrenia. In this study, Korean patients with schizophrenia showed impairments in the recognition of sad, fearful, and angry faces, but they were as accurate as controls in the recognition of happy emotions. Patients with schizophrenia also exhibited different patterns of misidentification relative to normal controls. These findings were consistent with previous studies carried out with different ethnic groups, suggesting cross-cultural similarities in the impairment of facial recognition in schizophrenia.
Schizophrenic patients in this study were found to be less accurate in the recognition of sad, fearful, and angry facial expressions. Note that these results were obtained regardless of the inherent difficulties of the given emotion. For example, the correct response rate for sad expressions was 93%, whereas that for fearful expressions was just below 50% even in normal controls. These rates corroborated our initial validation study
As to our result showing a trend toward an association between response time and negative symptoms, delayed response times in schizophrenia can be accounted for by many confounding factors such as negative symptoms, persistent illness, medication induced movement disorders, and impaired general cognitive functioning.
With regard to the correlations between recognition and symptoms, worse performance on the facial emotion recognition test was associated with more severe symptoms, exhibiting a stronger relationship with negative symptom scores. A large body of previous literature, mostly using the Scale for the Assessment of Negative Symptoms, has indicated that negative symptoms affect the performance of facial emotion recognition in schizophrenic patients.
In this study, schizophrenic patients misidentified neutral cues as emotional, showing a negative as well as positive bias. Kohler et al.
This study has some limitations. Impaired emotion recognition in this study was limited to a group of patients with schizophrenia who were clinically stable yet chronic. The effect of medication was not controlled in this study. Previous findings regarding this issue have been inconsistent.
In conclusion, using Korean emotional faces, we found that Korean patients with schizophrenia showed impairments in emotion recognition, specifically for sad, fearful, and angry faces. They also exhibited differences in the pattern of error rates. Previous studies with other ethnic groups as well as this study suggest cross-cultural similarities in the deficit in facial recognition in schizophrenia.
Recognition of facial emotions by patient with schizophrenia and normal controls. *p<0.05, **p<0.01.
Demographic and clinical data for schizophrenia and control groups
Standard deviations appear in parentheses. *p<0.001. PANSS: Positive and Negative Syndrome Scale
Response time (second)
Relationships between psychiatric symptoms and correct response of facial recognition test within schizophrenia group (N=55)
*p<0.05, **p<0.01
Relationships between psychiatric symptoms and response time of facial recognition test within schizophrenia group (N=55)
*p<0.05, **p<0.01
Error profile of patients with schizophrenia and normal controls for each facial emotion
*†‡the groups did not differ significantly (p=0.702 for happiness; p=0.415 for sadness; p=0.315 for fear by Fisher's exact test), §∥¶**the groups differed significantly (p=0.014 for anger; p<0.001 for surprise; p=0.012 for disgust; p<0.001 for neutral by Fisher's exact test)