Associations of emotion recognition, loneliness, and social functioning in euthymic patients with bipolar disorder

Abstract

Emotion recognition deficit is related to impaired community functioning. Loneliness is also associated with impaired social performance. However, the way in which emotion recognition and loneliness may contribute to social functioning remains unclear in euthymic patients with bipolar disorder. We aimed to examine emotion recognition ability in Han Chinese euBD patients relative to healthy controls (HCs) and to investigate the associations between emotion recognition, loneliness, and social functioning. Thirty‐nine HCs and 46 euthymic BD patients completed an emotion recognition task and nonsocial cognitive measures related to executive function and attention. The UCLA loneliness scale and Social Performance Scale were administered to evaluate psychological loneliness and social functioning, respectively. We observed lower emotion recognition accuracy, higher loneliness, and poorer social functioning in the BD patients after adjustment for demographic data. Loneliness was negatively associated with global social functioning in both the BD and HC groups. Higher loneliness and lower emotion recognition accuracy were associated with poorer social functioning in euthymic BD in different subdomains. Our study confirmed a subtle impairment of emotion recognition ability in euthymic BD. Loneliness impacts globally on social functioning, while emotion recognition ability may affect specific subdomains of social functioning in euthymic BD. Alleviation of loneliness and enhancement of social cognition might improve social functioning in BD patients.

INTRODUCTION

Bipolar disorder (BD) is a chronic psychiatric illness characterized by recurrent manic and depressive symptoms and is associated with significant psychosocial disability in terms of work performance, family disturbance, and social dysfunction. Subsyndromal residual symptoms and cognitive impairments in the euthymic phase appear to be relevant to low functioning.^1,2

Mounting evidence suggests the existence of persistent neurocognitive deficits in euthymic BD patients.³ The most affected domains are attention, verbal learning, and memory, in addition to executive functions. Deficits in social cognition, a multi-dimensional construct that encompasses cognitive processes underlying social behaviors,⁴ are also observed in euthymic BD patients. The presence of neurocognitive and social cognitive deficits that are independent from mood changes represents a trait feature of BD, and may be associated with persistent functional impairments observed in the euthymic state.⁵ Furthermore, social cognition seems to mediate the relationship between neurocognition and community functioning.⁶

Impaired emotion recognition, whether in the euthymic or symptomatic state, is one of the most prominent social cognitive deficits in BD. However, most available data are from Anglo-Saxon populations. Few studies focused on social cognition of BD patients in other ethnicities, such as Han Chinese, an East Asian ethnic group that comprises more than 95% of the population in Taiwan. The existing evidence is controversial in the euthymic state of BD. Pan et al.⁷ revealed that manic BD patients performed significantly poorer in recognizing negative emotions; however, the performance of euthymic BD patients in recognizing emotional faces was only numerically worse than that of the controls. In contrast, Liu et al.⁸ showed that euthymic Han Chinese BD patients exhibited significant social cognitive deficits in terms of understanding emotion, but no statistical difference in perceiving emotional faces was observed as compared with controls.

Social functioning can be assessed in various domains, including occupation, interpersonal relationships, leisure activities, and autonomy. Emotion recognition, as the basis of social cognitive function, is an important predictor of social functioning, as the ability to accurately identify and interpret emotional expressions in other people is crucial to social interaction. For example, in schizophrenia patients, emotion recognition has been identified as an independent correlate of social functioning.⁹ Similar associations between multiple social cognitive domains and social functioning have also been observed in symptomatic and euthymic BD patients. Poorer performances in tasks assessing the understanding and management of emotions were found to be related to greater social functioning impairment in BD patients across all stages.⁵

Loneliness, one of the subjective measures of psychological wellbeing, has been repeatedly demonstrated to be associated with poor social functioning and increased risks of developing a variety of physical health and mental problems.^10,11 Loneliness has been defined as the absence of socially integrative relationships and of intimate relationships.¹² Previous studies revealed that lower perceived social support was found to be a significant predictor of greater impairment in functioning in BD,¹² but less literature has directly reported an association between loneliness and functional impairments in this population. Moreover, reports of the association between loneliness and social cognition are inconsistent.^13,14 Thus, the relationships between objective social cognitive deficits, subjective loneliness and social functioning remain unclear in euthymic BD.

Owing to the inconsistent data regarding emotion recognition ability in euthymic Han Chinese BD patients in a Taiwanese sample in the current literature, the current study aimed to reassess the ability in this population. We hypothesized that euthymic BD patients would have a poorer emotion recognition accuracy as compared with healthy subjects^15–17 in a Taiwanese population. Further, we aimed to explore the relationships between emotion recognition, loneliness, and social functioning in euthymic BD patients of a Han Chinese population. We hypothesized that both greater subjective loneliness and poorer objective emotion recognition accuracy would be associated with poorer social functioning. As evidence has shown that social cognition is a mediator between neurocognition and social functioning in first episode psychosis (including BD with psychotic symptoms),¹⁸ we also examined its relationships with executive function and attention, which are two of the most commonly impaired areas of cognitive function in BD patients.

MATERIALS AND METHODS Ethics statement

The Institutional Review Board for the Protection of Human Subjects at National Cheng Kung University Hospital approved the research protocol (B-BR-105-086). Patients were recruited via referrals from the psychiatric outpatient department of National Cheng Kung University Hospital, and healthy subjects were recruited via the internet and public advertisements. All participants signed written informed consent forms after the procedures had been fully explained.

Subjects

A total of 46 BD patients (BD I: 26, BD II: 20) and 39 healthy controls (HCs) were enrolled in the study. The subjects partly overlapped with those in Liu et al.⁸ Subjects underwent the diagnostic Mini International Neuropsychiatric Interview (MINI), and rating scales were completed by trained psychiatrists, including the 17-item Hamilton Depression Rating Scale (HDRS) and the 11-item Young Mania Rating Scale (YMRS). The inclusion criteria were: (i) aged 18–65 years; (ii) patients were diagnosed with BD according to the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5) criteria; (iii) patients were in a euthymic state, which was defined as a HDRS total score lower than 8 and a total YMRS score lower than 8.

Thirty-nine HCs were recruited as the control group. The inclusion criteria for the control group were: (i) aged 18–65 years; (ii) participants must be without a history of neuropsychiatric diseases or major physical illnesses (such as intellectual disabilities, autism spectrum disorder, BD, major depressive disorders, psychotic disorders, substance dependence, epilepsy, or severe head trauma).

Any subjects who met the one or more of the following criteria were excluded: (i) a serious surgical condition or physical illness; (ii) patients who were pregnant or breastfeeding; (iii) a DSM-5 diagnosis of substance abuse within the past 3 months; (iv) previous use of any psychotropic agent in healthy subjects; (v) an organic mental disease, mental retardation, or dementia.

Social cognitive and psychological measures Diagnostic Analysis of Non-Verbal Accuracy 2-Taiwan version

The Diagnostic Analysis of Non-Verbal Accuracy 2-Taiwan version (DANVA-2-TW)^19,20 is a computerized measure and consists of 60 facial photographs and 60 voice clips representing specific emotions and intensities, including 12 for each of the four basic emotion categories (happy, sad, angry, and fearful) and 12 neutral stimuli. In the present study, neutral stimuli were not included in the analysis. The facial photos were displayed on a computer screen in the full-screen mode with a resolution of 1024 × 768, and voice clips were played on a computer and delivered to the participants through an earphone. The faces and voices were presented in different sessions, and the participants were asked to make a forced choice from the four emotional categories. The accuracy values were the ratios of correctly answered items per emotion category, ranging from 0 (completely inaccurate) to 1 (completely accurate). The overall accuracy of facial or prosodic emotion recognition was the average of all emotions of the same sensory modality. The total accuracy of nonverbal emotion recognition was derived by averaging the overall facial and overall prosodic accuracy. To account for response bias, we computed the corrected accuracy for each emotion according to the method outlined by Tseng et al.²¹

University of California, Los Angeles loneliness scale

The Chinese version of the University of California, Los Angeles (UCLA) loneliness scale^22,23 is a 20-item scale that measures one's subjective feelings of loneliness and social isolation. Participants rate each item on a four-point Likert scale from 1 (never) to 4 (often), and individuals with higher values are those with more subjective feelings of loneliness.

Self-Reported Graphic version of the Personal and Social Performance scale

The Self-Reported Graphic version of the Personal and Social Performance scale (SRG-PSP) scale was designed following the four PSP domains developed by Morosini et al.²⁴: socially useful activities, personal and social relationships, self-care, and disturbing and aggressive behavior. It is a self-reported questionnaire composed of 22 items showing cartoon-like pictures. Each item is rated on a scale from 1 (never) to 3 (always), and the domain score was summed. The global score was a summation of the scores of the socially useful activities, personal and social relationships, and self-care domains, and the reversed score of the disturbing and aggressive behavior domain. A higher global score indicated a higher level of personal and social functioning.

Wisconsin Card-Sorting Test

The Wisconsin Card-Sorting Test (WCST) was conducted by an experienced clinical neuropsychologist. There were 64 cards in the test. All definitions of the indices were as described in the WCST manual.²⁵ Using a computerized version of the WCST, the patients were required to match response cards to four stimulus cards along one of three dimensions (color, form, or number) on the basis of sign feedback (correct or wrong). The subjects were not given any information about the dimensions. After sorting a series of 10 cards in one category, the subject was asked to sort the cards again in a different category. The indexes of categories completed and preservative errors were used to assess performance in the WCST.²⁶

Continuous Performance Test

The Continuous Performance Test (CPT) is a psychological test for humans that primarily measures attention.²⁷ Only the AX task (in which subjects were asked to respond whenever the number “9” was preceded by the number “1”) was used in the present study. Each test session began with a 2-min practice period (repeated if the subjects required it) to ensure that the subject knew how to press the button correctly. During the test, numbers from 0 to 9 were randomly presented for 50 ms each, at a rate of one per second. A total of 331 trials, 34 (10%) of which were target stimuli, were presented over 5 min in each session. In this study, subject responses were recorded automatically on a diskette using the CPT machine (Sunrise Systems, version 2.20, Pembroke, MA).²⁸ The rater monitored each subject's performance through the computer monitor. A higher attention test index (d′) indicates a better processing capability.

Statistical analyses

We analyzed the data using the Statistical Package for Social Sciences 22.0 for Windows (SPSS Inc., Chicago, IL). Categorical variables were expressed as numbers and percentages, and continuous variables as means ± SDs. A p value <0.05, two-tailed, was considered statistically significant. Because the level of education was significantly different between groups, analysis of covariance (ANCOVA) was used to test the group differences, controlling for education. Considering that age, education level in years and sex were the best predictors of emotion recognition performance in a normative sample using multiple regression,^29,30 adjusted scores of the DANVA-2-TW were also calculated based on data of HCs collected in Tseng et al.,³¹ corrected for age, sex, and education level in years. For the loneliness scale and SRG-PSP scale, we did not apply adjusted scores, as the results regarding the predictive effect were inconsistent.

As both social cognitive ability and subjective psychological factors are crucial factors in social functioning, we used these variables as independent variables in a linear regression analysis to examine the significant predictive factors of the SRG-PSP scale and subscale scores. Covariates of neurocognitive measures, including executive function and attention, as well as years of education, were entered into the regression analysis. The interactions between group and social cognitive ability/subjective psychological factors on the SRG-PSP sale and its subscales were also analyzed. Regarding analyses of the subscales of the SRG-PSP scale, the level of significance was adjusted to p <0.01 considering the issue of multiple comparisons when the regression models were performed.

RESULTS Demographic data and group differences

The demographic, social demographic and clinical characteristics of the BD patients and HCs are summarized in Table 1.

TABLE 1 Group comparisons between the patients with euthymic BD and healthy controls

	Healthy controls			BD patients			Statistics
	N	Mean	SD	N	Mean	SD	t/χ2	p	Fa	pa
Sex (M/F)	17/22			19/27			0.05	0.832
Age	39	33.41	11.22	46	37.11	13.50	−1.36	0.178
Educational years	39	17.03	2.50	46	14.57	2.06	4.98	<0.001
UCLA loneliness score	39	37.59	8.23	39	46.95	10.72	−4.32	<0.001	12.55	0.001
YMRS total score	39	0.00	0.00	39	0.85	1.53	−3.45	0.001	13.28	<0.001
HDRS total score	39	0.54	0.94	39	1.74	1.90	−3.55	0.001	7.86	0.006
WCST perseverative errors	34	7.44	4.47	44	11.95	10.15	−2.42	0.018	0.17	0.684
WCST completed categories	34	3.88	1.27	44	2.75	1.71	3.22	0.002	0.90	0.347
CPT unmasked d′	34	4.71	0.26	44	4.23	0.75	3.59	0.001	2.02	0.159
CPT mask d′	34	3.89	1.01	37	3.40	1.18	1.86	0.067	0.74	0.392
DANVA-2-TW
Total	39	0.71	0.11	46	0.61	0.17	3.30	0.001	2.48	0.119
Face total	39	0.66	0.14	46	0.55	0.17	3.06	0.003	1.80	0.184
Prosodic total	39	0.76	0.12	46	0.66	0.20	2.93	0.004	2.00	0.161
Emotional categories
Happy	39	0.83	0.08	46	0.73	0.19	2.90	0.005	2.62	0.110
Sad	39	0.68	0.13	46	0.59	0.19	2.43	0.017	0.91	0.343
Angry	39	0.66	0.12	46	0.57	0.17	2.95	0.004	1.40	0.241
Fearful	39	0.68	0.16	46	0.53	0.22	3.38	0.001	2.57	0.113
Adjusted DANVA-2-TWb
Total	39	0.67	0.11	46	0.60	0.14	2.18	0.033
Face total	39	0.62	0.14	46	0.55	0.16	2.38	0.020
Prosodic total	39	0.71	0.13	46	0.66	0.17	1.56	0.123
Emotional categories
Happy	39	0.77	0.09	46	0.71	0.17	1.86	0.067
Sad	39	0.59	0.13	46	0.56	0.17	0.90	0.369
Angry	39	0.62	0.11	46	0.56	0.15	2.04	0.044
Fearful	39	0.66	0.17	46	0.56	0.20	2.58	0.012
SRG-PSP
Global score	39	39.59	3.47	40	34.7	6.59	4.11	<0.001	8.47	0.005
Socially useful activities	39	16.15	1.86	41	14.24	2.61	3.76	<0.001	8.26	0.005
Personal and social relationships	39	12.1	2.14	41	9.46	3.03	4.48	<0.001	11.18	0.001
Disturbing and aggressive behavior	39	5.31	0.66	41	5.78	1.24	−2.12	0.037	1.67	0.201
Self-care	39	16.64	1.48	42	16.62	1.99	0.06	0.955	0.28	0.596
Religion (with/without)		24/	15		29/	11	1.08	0.300
Participation in community	39	5.79	3.82	40	5.43	5.28	0.36	0.723
Quality of friendships	39	5.13	3.34	39	5.23	4.16	−0.12	0.905
Available social resources	39	21.31	6.26	33	19.42	5.32	1.36	0.178
Marital status (n)							3.01	0.222
Single		28			24
Married		10			11
Separated		0			0
Divorced		1			5
Widowed		0			0
Occupation (n)							10.60	0.060
Unemployed		0			4
Homemaker		1			2
Part-time job		1			4
Full-time job or student		35			25
Volunteer		1			0
Retired		1			4
Onset age				45	28.37	12.13
Duration of illness				45	8.72	6.44
Inpatient (yes/no)					24/	22
Number of admissions in recent 10 years				46	1.30	1.71

Abbreviations: BD, bipolar disorder; CPT, Continuous Performance Test; DANVA-2-TW, Diagnostic Analysis of Non-Verbal Accuracy 2-Taiwan version; HDRS, 17-item Hamilton Depression Rating Scale; SRG-PSP, Self-Reported Graphic version of the Personal and Social Performance scale; YMRS, 11-item Young Mania Rating Scale; WCST, Wisconsin Card-Sorting Test.

^aAge, sex, and educational years were controlled.

^bAge, education in years, and sex were adjusted according to performance of healthy subjects in Tseng et al.³¹

The two groups were comparable in terms of age, sex, marital status, occupational status, and available social resources. The euthymic BD patients had fewer years of education (p <0.001). With regard to clinical symptoms, depressive and manic symptoms, the BD patients presented higher subsyndromal depressive (HDRS = 1.74 ± 1.90, p = 0.001) and manic (YMRS = 0.85 ± 1.53, p = 0.001) symptoms. Concerning psychological factors and social functioning, the BD group had higher scores on the UCLA loneliness scale and lower scores on the SRG-PSP scale.

Regarding cognitive function, the BD group exhibited poorer performances in social cognitive measures (DANVA-2-TW total score, emotional face subtest and prosodic voice subtest) and selected neurocognitive measures (WCST categories completed and perseveration errors; CPT d′). These differences became nonsignificant after controlling for age, sex, and level of education (Table 1). In order to evaluate the emotion recognition ability of the euthymic BD patients more accurately, we calculated the adjusted score according to a larger HC sample,³¹ which again showed a significantly lower emotion recognition ability in the euthymic BD group (t = −2.18, p = 0.033), mainly in the facial expression subtests (t = −2.38, p = 0.020), anger emotion (t = −2.04, p = 0.044), and fearful emotion (t = −2.58, p = 0.012).

There were no significant differences between the BD I group and BD II group in the abovementioned measures (data not shown).

Correlation and regression analyses

In a subgroup of participants who completed all the neurocognitive, social cognitive, and social functioning measures, we explored the associations of social cognitive measures, loneliness, and social functioning. As neurocognitive measures (WCST perseveration and CPT unmasked d′) and years of education were significantly different between groups, these variables were also entered as predicting variables in the following regression analyses to examine the major contributing factors to social functioning (SRG-PSP scale and its subscales) in each group.

In the BD group (n = 34), the UCLA loneliness scale total score was negatively associated with the SRG-PSP global score (β = −0.52, p = 0.003; p of interaction between group and UCLA loneliness scale total score on SRG-PSP global score = 0.20). We did not observe any correlations between neurocognitive and social cognitive measures (CPT, WCST, and DANVA-2-TW) and the global score of the SRG-PSP scale. When examining associations with the subscales of the SRG-PSP scale, we observed a significant negative association between the UCLA loneliness scale total score and the personal and social relationships subscale (β = −0.61, p <0.001; p of interaction between group and UCLA loneliness scale total score on personal and social relationships subscale = 0.16), while the DANVA-2-TW total score was associated with a better score on the self-care subscale (β = 0.68, p = 0.002; p of interaction between group and DANVA-2-TW on self-care subscale = 0.014) (Figure 1). In the HC group (n = 34), higher loneliness (β = −0.47, p = 0.008; ps of interaction between group and UCLA loneliness scale total score on the SRG-PSP global score = 0.20) and a lower perseveration error (β = 0.36, p = 0.049; p of interaction between group and perseveration error on SRG-PSP global score = 0.024) were associated with poorer social function. Emotion recognition was not associated with the global score or any subscale score of the SRG-PSP scale (Table 2). These analyses were repeated with the inclusion of age and sex as additional variables, and similar results were obtained (data not shown).

FIGURE 1. Scatter plots of (A) UCLA loneliness score and SRG-PSP global score and (B) DANVA-2-TW total score and SRG-PSP self-care score. BD, bipolar disorder; DANVA-2-TW, Diagnostic Analysis of Non-Verbal Accuracy 2-Taiwan version; SRG-PSP, Self-Reported Graphic version of the Personal and Social Performance scale

TABLE 2 Regression analysis examining predictive factors of the SRG-PSP scale and its subscales

		Global score		Socially useful activities		Personal and social relationships		Disturbing and aggressive behavior		Self-care
		Beta	p	Beta	p	Beta	p	Beta	p	Beta	p
Healthy controls	(Constant)		0.004		0.091		0.465		0.330		0.001
	UCLA loneliness score	−0.47	0.008	−0.14	0.428	−0.40	0.028	0.06	0.747	−0.30	0.119
	DANVA-2-TW total score	0.14	0.446	0.16	0.435	0.21	0.288	−0.01	0.976	−0.17	0.420
	WCST perseverative errors	0.36	0.049	0.36	0.065	0.40	0.036	0.34	0.087	−0.09	0.652
	CPT unmasked d′	0.07	0.711	−0.02	0.917	0.20	0.294	0.32	0.106	0.00	0.986
	Educational years	0.15	0.381	0.28	0.136	−0.07	0.706	0.05	0.772	0.10	0.588
BD patients	(Constant)		<0.001		0.001		0.041		0.082		<0.001
	UCLA loneliness score	−0.52	0.003	−0.32	0.097	−0.61	<0.001	0.32	0.084	−0.19	0.232
	DANVA-2-TW total score	0.21	0.304	0.00	0.999	−0.08	0.667	−0.28	0.235	0.68	0.002
	WCST perseverative errors	−0.28	0.135	−0.23	0.297	−0.14	0.431	0.19	0.367	−0.33	0.080
	CPT unmask d′	−0.12	0.533	−0.08	0.730	0.14	0.442	0.02	0.935	−0.51	0.011
	Educational years	0.01	0.977	−0.04	0.864	0.20	0.314	0.09	0.715	−0.19	0.364

Abbreviations: CPT: Continuous Performance Test; DANVA-2-TW: Diagnostic Analysis of Non-Verbal Accuracy 2-Taiwan version; SRG-PSP: Self-Reported Graphic version of the Personal and Social Performance scale; WCST: Wisconsin Card-Sorting Test.

DISCUSSION

The present study was conducted to reassess the emotion recognition ability of euthymic Han Chinese BD patients, and to clarify the associations between subjective loneliness, objective emotion recognition ability, and social functioning. We observed that the euthymic BD patients performed worse in the emotion recognition task after adjustment for age and educational level, in line with results reported in the majority of western literature. The euthymic BD patients had a greater degree of loneliness and poorer personal and social performance, suggesting that even in the euthymic state, BD patients still suffer from significant functional impairments and poorer psychological wellbeing.

In contrast to previous studies by Pan et al.⁷ (which employed the DANVA-2-TW face subtest) and Liu et al.⁸ (MSCEIT, perceiving emotion branches), in which prominent emotion recognition deficits were not observed in euthymic Han Chinese BD patients, we observed a significantly lower accuracy in euthymic BD patients when the scores were carefully adjusted according to a larger healthy sample enrolled in our previous study.³¹ Education level is generally lower in most clinical populations with severe mental illnesses such as schizophrenia and BD, which suggests early involvement of pathological changes even before illness onset, and/or adverse effects of early events and environments. It has been a matter of constant debate whether matching of patients in terms of IQ or education level with HCs could result in a biased selection of patients with better overall functioning. As the social cognitive deficits in euthymic BD patients are subtle^32–34 as compared with the symptomatic phase, the use of education level-adjusted scores instead of directly controlling for years of education would be more sensitive for detection of a group difference.

Regarding the sensory modalities of the stimuli, as was observed in the current study, several studies have demonstrated a facial emotion processing deficit in euthymic BD patients,^35,36 despite the deficit being less prominent than in manic or depressive episodes. However, evidence regarding impairment of discriminating emotional prosody among euthymic BD patients is limited, and conflicting results have been reported.³⁷ Subtle disparities in task elements, such as the use of more than one actor, could introduce variability in pronunciation. We also speculated that cultural influence may impact more on prosodic emotion recognition and further increase the variability. Given the relatively small sample size of this study, the power might not be sufficient to detect such a difference.

The results of the current study also showed a greater degree of loneliness and poorer personal and social function in the euthymic BD patients, despite having a similar social demographic background to the HCs. Loneliness is often caused by a combination of maladaptive social cognitions, impaired social skills, lack of opportunities to participate in social activities, and functional challenges in patients with major mental illness.³⁸ As hypothesized, we observed a negative association between subjective loneliness feelings and social functioning in both the HC and BD groups, as has been observed in patients with depression.³⁹ Psychological loneliness seems to be a universal contributing factor to social functioning across diagnoses including BD, even in the euthymic phase.

We hypothesized that both objective emotion recognition ability and subjective psychological loneliness would be associated with social functioning in euthymic BD. Our results at least partially supported the hypothesis. We observed that the DANVA-2-TW total score was positively associated with the self-care subscale of the SRG-PSP scale. Furthermore, in the adjusted regression model, the positive association between the adjusted DANVA-2-TW total score and the SRG-PSP scale total score was significant. In contrast, the ULCA loneliness scale score was negatively associated with the SRG-PSP scale total score and personal and social relationships. Our results suggested that objective emotional recognition function and subjective psychological wellbeing may contribute to different domains of social functioning in euthymic BD. Social cognition positively contributed more to the instrumental level, and loneliness contributed negatively to the interpersonal level of social functioning.

Surprisingly, we did not observe a significant association between DANVA-2-TW and the UCLA loneliness scale. A possible explanation is that the association between emotion recognition and loneliness may be emotion-specific.⁴⁰ Loneliness is associated with poorer recognition of fear, but better recognition of prosocial emotions like friendliness in adolescents. While emotion-specific analyses may help to understand the possible associations, our sample size limited the power to perform such sophisticated analyses.

Although most studies have highlighted an effect of subsyndromal depressive symptoms on the link between social cognition and social functioning, several studies have failed to identify one. In our study, although the BD patients presented a significantly higher level of subsyndromal depressive symptoms, the HAMD scores were too low to be clinically significant.

As far as we are aware, this was the first study reporting an association of emotion recognition and social functioning in euthymic BD patients in a Han Chinese population. We used a comprehensive, dual-modality, cultural-friendly instrument for the measurement of emotion recognition, and considered subjective loneliness as an additional important contributing factor to social functioning.

There were several limitations of the current study. First, a difference in education level existed between groups, a general limitation in the research field of major psychiatric disorders owing to the disruptive nature of these illnesses. We used adjusted scores to alleviate the negative impact on the statistical power in our main measurement, that is, DANVA-2-TW. Larger sample size in future studies would allow more robust statistical control of the possible influences on all measures. Further, a larger sample would also enable emotion-specific analyses to clarify the association between emotion recognition and loneliness. Second, the study was cross-sectional, and strong inferences of causal relationships were not possible. A longitudinal design would provide greater insight into the effects of emotion recognition ability and psychological well-being on social functioning. Third, emotion recognition is the most basic component of social cognition. Generalization of the results to present the overall effect of social cognition on social functioning needs to be performed cautiously. Fourth, not all participants completed all the neuropsychological and psychosocial measures, particularly the BD patient group. Reasons for this were the late introduction of the SRG-PSP scale into the study, and the patients being less tolerable of the relatively long study schedule. Again, generalization of the association reported should be cautious.

In conclusion, our study demonstrated subtle emotion recognition deficits in euthymic bipolar patients when the influence of education level was carefully considered. We also demonstrated that emotion recognition ability and psychological loneliness were associated with different domains of social functioning in opposite directions. Longitudinal research is required to confirm the causal relationships between emotion recognition ability, loneliness, and social functioning, with a larger sample size and adoption of other social cognitive measures.

ACKNOWLEDGMENTS

The authors are indebted to the research participants.

CONFLICT OF INTEREST

The authors declare no conflict of interest.