1. Introduction
Burnout, defined as “a syndrome conceptualized as resulting from chronic workplace stress that has not been successfully managed” [
1], has been a major occupational health issue for many years. A systematic review conducted by Salvagioni et al. [
2] revealed that burnout led to insomnia, depressive symptoms, fatigue, coronary heart disease, headaches, and other psychological and physical symptoms. Although burnout is not classified as a disease [
1], it may lead to many psychological and physical symptoms or illnesses. Thus, it is important for employers and supervisors to manage their subordinates’ levels of burnout to protect their health.
Burnout has previously been studied in human service workers in terms of emotional labor. However, Leiter and Schaufeli [
3] pointed out that human service workers’ burnout is only one form experienced in various occupations, and burnout research is now being conducted for a variety of workers. In fact, burnout studies have been conducted for forestry workers [
4], electronics company workers [
5], and other occupations.
Burnout consists of three components: emotional exhaustion, depersonalization, and reduced personal accomplishment [
6]. Emotional exhaustion is feelings of energy depletion or exhaustion due to prolonged exposure to high job stressors. Depersonalization refers to the state in which one responds to others in an unkind and inhuman manner, having exhausted their emotional resources. Reduced personal accomplishment is the feeling of being less competent and less accomplished due to a sharp decline in performance. Among these three components, Maslach et al. [
7] stated that “exhaustion is the central quality of burnout and the most obvious manifestation of this complex syndrome.” Thus, emotional exhaustion may be a better predictor of an employee’s burnout. Employees who feel burnout are exhausted both physically and psychologically, and are in a state of “emotional numbness” [
8].
Positive emotions can broaden and build our thinking and behavior [
9,
10], which may lead to increased self-efficacy [
11] and work engagement [
12]. For example, Rogala [
13] found that positive emotions increase job crafting, meaning ingenuity in one’s work [
14], through a longitudinal study. Job crafting is strongly associated with self-efficacy [
13] and work engagement [
15]. Increasing job crafting by improving positive emotions may lead to a positive attitude toward work without giving up, resulting in an increased possibility of success at work and less emotional exhaustion.
Recently, Artificial Intelligence (AI) technology has advanced to detect our emotions from facial expressions. Several commercial emotion cognition systems have been developed, such as Amazon Rekognition, Baidu Research, Affectiva, and Microsoft Azure [
16]. Most of the software can detect the type and percentage of each emotion in our facial expressions. Measuring emotions with questionnaires may introduce biases such as social desirability. However, if the emotions an individual is feeling can be detected from their facial expressions, burnout may be better detected.
The purpose of this study is to examine the relationship between the emotions detected by the emotion cognition system and burnout among workers. We hypothesized that burnout survivors, especially those who feel high levels of emotional exhaustion, feel wear and tear on their emotional resources due to chronic job stressors and are, therefore, less likely to express their emotions in their facial expressions.
4. Discussion
This study examined the prospective relationships between emotions and burnout. One hundred and forty-one workers at an IT products and services trading company were asked to take facial images when they started and left work for three months and responded to a burnout questionnaire once a month. Correlation analysis revealed that the neutral emotion in Period 1 was significantly and positively correlated with burnout at Time 2, and the neutral emotion in Period 2 was also significantly and positively correlated with burnout at Time 2 and Time 3. Happiness in Period 1 was significantly and negatively correlated with burnout at all timepoints, and surprise in Period 1 was significantly and positively correlated with burnout at all timepoints. Hierarchical multiple regression analyses revealed that happiness in Period 1 was significantly and negatively associated with burnout at Time 2. This association was also observed after the various covariates were included. However, burnout at Time 3 was not significantly related to any emotions in Period 1. These results partially support the hypothesis that burnout survivors are less likely to express their emotions in facial expressions.
Hierarchical multiple regression analysis suggests that even after controlling for the effects of various covariates, the low expression of happiness may predict burnout immediately afterward. The mean score for happiness, as expressed over one month, ranged from 0.5824 to 0.0000, indicating that happiness is the most likely emotion to be expressed, except for the neutral emotion, by workers when starting work. Several previous studies showed a negative association between happiness and burnout. Sharif et al. [
19] found a significant negative association between happiness and burnout in a study of 344 nurses working in hospitals in Iran. Similar associations were found in a study of 1147 European pilots [
20] and a study of 548 German general practitioners [
21]. Most of these studies were cross-sectional and administered questionnaires measuring burnout and happiness; thus, common method bias [
22] may have occurred in these studies because they used the same measurement method. However, our study found the same association between emotions as detected by the face cognition system and burnout as measured by the questionnaire. Therefore, our results may corroborate the negative association between happiness and burnout. To the best of our knowledge, no previous study has reported a prospective association between happiness and burnout; our results may suggest that happiness, as detected by the face cognition system, may be an indicator predicting burnout immediately afterwards.
Although the results of the correlation analysis showed a positive association between neutral expression and burnout, it could not be entered as an independent variable in the hierarchical multiple regression analysis since multicollinearity was observed for the neutral emotion. While Microsoft Azure is considered to be able to recognize emotions relatively more accurately than other face cognition systems [
16,
23], it is recognized that it also detects many neutral emotions [
16]. Since the maximum detection range for neutral emotions in this study was extremely high, multicollinearity was more likely to be observed when other emotions were included as independent variables. Thus, it is necessary to reexamine the relationship with burnout using other face cognition systems in the future.
Several limitations of this study should be noted. First, the survey was conducted on employees working for one Japanese IT products and services trading company, but 71.0% of the participants had roles other than IT engineers, such as planning, purchasing, and sales. Therefore, there is a possibility that these results can be applied to other organizations, although further study should be warranted. Second, the frequency at which facial images were taken varied between participants. In this study, since we did not force participants to take pictures, the number of shots that were taken could not be kept constant. Although the average score for each emotion for each participant over one month was used as an indicator, it will be necessary to conduct continuous filming to examine the relationship between changes in emotions and burnout, since the expression of emotions changes from day to day. Third, data are missing for those who did not participate in the study. Those who had burned out during or before participation were not included in this study, and this study may not have shown the real association between emotions and burnout.
Finally, we offer some practical implications from this study. Since burnout may occur within one month if the expression of happiness is low, supervisors should consider this as a state in which burnout risk is increasing in their subordinates. Active monitoring and early detection of signs of burnout, and expressing less happiness, may give more opportunities for supervisors to take care of their subordinates.