Conceived and designed the experiments: LRD AA. Performed the experiments: LRD. Analyzed the data: LRD. Contributed reagents/materials/analysis tools: LRD. Wrote the paper: LRD. Improvement of the manuscript: RK AA JAdB.
The authors have declared that no competing interests exist.
Recognition of others' emotions is an important aspect of interpersonal communication. In major depression, a significant emotion recognition impairment has been reported. It remains unclear whether the ability to recognize emotion from facial expressions is also impaired in anxiety disorders. There is a need to review and integrate the published literature on emotional expression recognition in anxiety disorders and major depression.
A detailed literature search was used to identify studies on explicit emotion recognition in patients with anxiety disorders and major depression compared to healthy participants. Eighteen studies provided sufficient information to be included. The differences on emotion recognition impairment between patients and controls (Cohen's d) with corresponding confidence intervals were computed for each study. Over all studies, adults with anxiety disorders had a significant impairment in emotion recognition (d = −0.35). In children with anxiety disorders no significant impairment of emotion recognition was found (d = −0.03). Major depression was associated with an even larger impairment in recognition of facial expressions of emotion (d = −0.58).
Results from the current analysis support the hypothesis that adults with anxiety disorders or major depression both have a deficit in recognizing facial expression of emotions, and that this deficit is more pronounced in major depression than in anxiety.
The ability to identify and interpret facial expressions of emotion is essential in human communication and social interaction. Ekman and Friesen
Discrimination of emotion from facial expressions has been the focus of a number of psychological studies over the past decades, and was later complemented by neurobiological findings
Emotional dysfunctions (e.g., difficulty in understanding emotions, difficulty in changing how one feels) are related to poor social functioning and can be considered as important features of psychopathology
Over the last two decades, research regarding emotional facial expressions in anxiety and major depression has focused on two areas: attentional bias and the ability to recognize emotions, with much more consideration devoted to the former. Several types of cognitive bias have been described in social phobia, generalized anxiety disorder and panic disorder. Biases involve attention, judgment, interpretation, imagery and memory (for a review, see: Clark and McManus
Cognitive deficits, such as perception and attention as well as motivational deficits have been associated with depression
The aim of the present review was to determine, by conducting a comprehensive meta-analytical synthesis of previous studies, the magnitude of the impairment in facial emotion recognition accuracy associated with anxiety disorders and major depression. In children with anxiety disorders there is evidence suggesting an impairment of emotion recognition, however the findings are inconsistent. Because such an impairment may provide clues about the role of developmental processes in anxiety disorders, we examined the extent to which this impairment is present in children with anxiety. We hypothesized that there would be a small difference for anxiety patients when compared to healthy controls in emotion recognition as they do not have substantial cognitive deficits. In contrast, we expected to find a larger impairment in emotion recognition in depressed patients.
The Web of Science (ISI) and PubMed databases were searched for the period 1980–2009. The search was performed using ”anxiety” and ”depression” combined with ”emotion”, ”facial expressions”, ”recognition”, ”discrimination”, ”labeling” as search terms. Additionally, more specific terms were used such as ”social phobia” (SP), ”generalized anxiety disorder” (GAD), ”panic disorder” (PD), ”posttraumatic stress disorder” (PTSD) and ”major depressive disorder” (MDD).
173 studies on anxiety disorders and 208 studies on depression were identified. Subsequently, title and abstract of the articles were screened for possible inclusion in the analysis. The identified studies were included if they met the following criteria: the diagnosis of major depressive disorder or anxiety disorders was made according to the DSM-III or DSM-IV criteria. Secondly, each study had to deal with a group of adults or children experiencing major depression or anxiety disorders and a control group. Third, behavioral measures of emotional facial expressions discrimination or identification accuracy had to be reported with sufficient statistical information for the computation of the effect size (d-value). This implies that means and standard deviations, t-values or F-values and the relevant means, and exact p-values had to be reported. Lastly, only studies published in English were included.
In this study we included facial expression discrimination paradigms measuring emotion recognition accuracy employing presentation of schematic or photographed facial expressions depicting neutral, emotional or ambiguous expressions. These stimuli were either presented on a computer screen, or in a booklet. The duration of the stimulus presentation was either fixed or unrestrained. Participants had to label each facial expression or they had to judge whether a pair of faces expressed the same emotion are or not. Responses were given either by pressing a key or by selecting one of the cards. The accuracy was one of the measures in the emotion recognition paradigms.
For each study the effect size (Cohen's d) was calculated for the difference in emotion recognition performance between the patient group and the control group. The d was calculated as the difference between the two group means, patient group minus control group and divided by the pooled standard deviation
After computing the effect size for each study, a meta-analytic method was used (Comprehensive Meta-analysis program,
Publication bias was tested using the Duval and Tweedie's trim and fill, and by inspecting the funnel plot. The Duval and Tweedie's trim and fill is a nonparametric method which concerns a simple funnel plot-based method of testing and adjusting for publication bias in meta-analysis, by using the ranks of the absolute values of the observed effect sizes and the signs of those effect sizes around the global effect size
Out of 381 identified studies, twenty-eight studies were potentially eligible for inclusion based on screening of the title and abstract, and the full text version of each of these manuscripts was further evaluated. Six studies were excluded because of insufficient data needed to calculate the effect size
The results of meta-analysis of emotion recognition in anxiety disorders and major depression are displayed in
k |
|
95% CI | Z | p* | I2 | Q | p** | |
Anxiety - Children | 5 | −0.03 (0.08) | −0.30, 0.24 | −0.21 | 0.831 | 0.00 | 2.99 | 0.560 |
Anxiety - Adults | 5 | −0.35 (−0.51) | −0.61, −0.10 | −2.69 | 0.007 | 22.30 | 5.15 | 0.270 |
Depression - Adults | 8 | −0.58 (−0.42) | −0.79, −0.36 | −5.17 | 0.000 | 33.75 | 10.57 | 0.160 |
In adults with anxiety disorders the meta-analysis showed a significant impairment (p = 0.007) with a medium magnitude (d = −0.35, 95% confidence interval: −0.61 to −0.10) of facial emotion recognition in anxiety disorders. There was a low variability among the effect sizes (I2 = 22.30). Correcting this for publication bias still resulted in a robust estimated effect size of d = −0.51 (95% confidence interval: −0.73 to −0.29). The overall effect size was larger than if we ignored a possible publication bias. More specifically, Duval and Tweedie's trim and fill suggested that there may be two missing studies with a negative effect (impaired emotion recognition in anxiety disorders compared to controls).
The meta-analysis of studies on major depression showed significant impairment of emotion recognition with a medium overall effect size of d = −0.58 (p<0.001). There was a moderate variability among the effect sizes (I2 = 33.75, Q = 10.57, p = 0.16). Once corrected for publication bias the relationship between the impairment of emotion recognition and major depression remained robust, although this overall effect size was reduced (d = −0.42, 95% confidence interval: −0.62 to −0.23) relative to that estimated from the original data.
The purpose of this study was to investigate whether and to what extent anxiety disorders and major depression are associated with impaired recognition of emotion in others. The results from the current analysis support the hypothesis that adults with major depression or anxiety disorders have an impaired recognition of facial expression of emotion, as substantiated by the medium effect size. This effect was not observed in children with anxiety disorders.
The present meta-analysis shows that children with anxiety disorders do not have an overall emotion recognition deficit. Some studies
Furthermore, the present meta-analysis revealed a moderate impairment of facial emotion recognition in adults with anxiety disorders. The underlying mechanism for this impairment is unknown, but attentional biases might be involved. Indeed, an emotion specific impairment has been suggested in association with anxiety disorders. For example, Kessler et al.
With regard to major depression, we also found a moderate overall emotion recognition impairment. Thus, patients with major depression may be compromised in recognizing emotions of other people from facial expressions. This may contribute to social dysfunction, as it has been well established that emotion recognition contributes to proficient social functioning
Some studies have reported specific emotion recognition impairment in depression. For example, Leppanen et al.
Another factor which might contribute to the present result is illness severity. Almost all studies included in the present review, with the exception of Archer et al.
Emotion recognition has been shown to be related to social functioning
Factors which might influence our results are the small sample size, differences in stimulus material and task complexity, and duration of illness which may have an impact on the outcome of emotion recognition tasks and cognitive tests. A second limitation concerns the fact that there were differences in the degree of severity of the anxiety disorders, not all of them being clinical patients. Another limitation is that we could not distinguish between individual emotions, because most studies did not report adequately detailed data to permit such comparisons. It would be of interest to investigate whether patients with major depression are selectively more impaired for certain expressions (e.g., happy) than others (e.g., sad), as was suggested by Surguladze et al.
In summary, we reviewed behavioral studies indicating the relevance of facial expression of emotion to anxiety disorders and major depression. The present findings suggest a global deficit in recognition of different types of emotions, which was more pronounced in major depression than in anxiety disorders. These emotion recognition deficits may contribute to compromise social functioning in these disorders.
Characteristics of included studies on anxiety disorders.
(DOC)
Characteristics of included studies on major depressive disorder.
(DOC)