Relationship between perception of facial emotions and anxiety in clinical depression: does anxiety-related perception predict persistence of depression?

Abstract

Within the framework of interpersonal theories on depression, it was postulated 1) that an anxiety-related mood-congruent bias with respect to the perception of facial expressions could be demonstrated in clinically depressed patients; 2) that the perception of negative facial emotions would be associated with co-occurring anxiety levels rather than with depression, and 3) that the putative anxiety-related bias would predict the subsequent course of depression. Such relationships would support the possible causal role of negative biases for the persistence of depression.

Thirty-nine depressed patients (thirty-six patients met the criteria for major depression, two had a dysthymic disorder and one patient suffered from a cyclothymic disorder) were studied. The patients judged schematic faces with respect to the emotions they express (fear, happiness, anger, sadness, disgust, surprise, rejection and invitation) at admission (T0), and after 6 (T1) and 30 (T2) weeks. Severity of depression (BDI) and anxiety (SCL-90) were assessed at these three points.

We found considerable support for the first 2 hypotheses: a) The perception of negative emotions was related to anxiety but not to depression (at T0 this association was significant and at T1 and T2 tendencies were found); b) When the level of depression was controlled for, significant relationships remained (emerged) between anxiety and the perception of negative emotions at each of the three different time points; c) Anxiety and perception of negative emotions covaried within subjects when large changes in depression/anxiety were involved, i.e. after 30 weeks. This relationship disappeared when depression change was partialled out. The third hypothesis was not confirmed: The perception of negative emotions did not predict the course of depression.

Although a direct relationship with depression persistence and a negative bias in the perception of interaction-relevant stimuli (i.e. facial emotions) in anxious depressed patients could not be found, the existence of such anxiety-related negative bias forms indirect evidence for the notion that this negative bias may mediate rejective attitudes of others towards depressives and consequently may contribute to an unfavourable course of depression.

Introduction

Interpersonal processes are presumed to play a role in depression persistence. Deficient or problematic social behaviours of depression-prone persons may underlie depressogenic processes by eliciting negative reactions in others, which may result in withdrawal by family and friends (Lewinsohn, 1974, Coyne et al., 1990, Segrin and Abramson, 1994). A solid body of evidence supports rejection attitudes of others towards depressed subjects (Segrin, 1993a, Segrin and Abramson, 1994). However, which processes underlie the emergence of such negative reactions in others is unknown. According to Beck's model, depression is characterized by mood-congruent biases, which operate all aspects of processing, such as attention, reasoning, memory (Beck et al., 1979). The cognitive system “filters” environmental information through imposing pre-existing memory representations on this new information. It has frequently been suggested that such enhanced processing of negative cues plays a role in the persistence of depression (Gotlib and Hammen, 1992, Mathews and McLeod, 1994). One may argue that the depression-prone person's perception of how (significant) others communicate emotions may be relevant for depression onset and persistence: a negative bias in the perception of interaction-relevant stimuli may serve as a mediator of rejecting attitudes of significant others. Knowing that about 60% of human communication is nonverbal (Burgoon, 1985, Hall et al., 1995), it seems reasonable to assume that the interpretation of other's nonverbal behaviour may play a role in depressogenic interactions.

That non-verbal processes are involved in depressogenic interactions has been presumed by several authors (Argyle, 1978, Youngren and Lewinsohn, 1980, Gotlib and Robinson, 1982, Troisi et al., 1989, Bouhuys and Albersnagel, 1992, Segrin, 1993b, Segrin and Abramson, 1994, Bouhuys et al., 1995). Supportive evidence has for instance been found with respect to nonverbal behaviours that express involvement in an interaction. It has been demonstrated that the nonverbal expression of high involvement by patients and by the interviewing person predicted persistence of depression (Troisi et al., 1989Bouhuys and Albersnagel, 1992Bouhuys and Van den Hoofdakker, 1993Geerts et al., 1995). Moreover, Geerts et al. (1996)revealed that lack of attunement of such involvement behaviours between patients and a care-giver also predicted an unfavourable course of depression. They suggested that differences in information processing may underlie their findings. Hence, the appreciation of nonverbal signals is supposed to play a (causal) role in depression. In particular the face signals emotional states and regulates other's behaviour (Darwin, 1965, Ekman, 1992). Human facial displays, like all animal signals, are schematized for signal value in social interactions and reflect intentions of future behaviour, that are interpreted by the recipient. One may assume that correctly processing of emotional meaning of facial expressions is important for normal relationships. For other non-verbal behaviours such assumption has been proven. For instance, interactions between persons who both correctly interpret non-verbal behaviours are experienced as more meaningful and are accompanied with more mutual support (Hodgins and Zuckerman, 1990, Hall et al., 1995). We suggest that the negative bias in interpretation of facial expressions serves a mediating role in the negative course of interpersonal behaviour. Thus, negative interpretation of other's emotions may underlie the often reported rejection that depressives elicit in others (see Segrin, 1993a). In other words, negative interactions may be mediated by depressed patient's negative interpretations of other's facial expressions. We therefore feel that studying non-verbal emotion expression is relevant in research on depression onset and persistence. Most research on cognition and emotion has been performed on non-clinical populations using different versions of a Stroop task. In those designs putative mood-congruent stimuli are mostly negatively tuned words. We presumed that the appreciation of nonverbally expressed emotions would bear more relevance for depressogenic processes.

It is well known that anxiety often coincides with depression (Ormel et al., 1993, Pasnau and Bystritskt, 1994). In addition, high levels of co-occurring anxiety predict an unfavourable course of depression (Kupfer and Spiker, 1981McLeod et al., 1992Coryell et al., 1992Ball et al., 1994). In perceiving and interpreting facial expressions at least two cognitive processes are involved: attention and interpretation. In their review on cognitions and emotions, Mathews and McLeod (1994)concluded that there is relatively little evidence to suggest that elevated levels of depression are associated with the facilitative processing of emotionally negative information. On the other hand they found considerable empirical support for the hypothesis that such encoding bias is associated with elevated levels of anxiety. The same could be concluded for interpretation of information: An interpretation bias favouring negative meanings was related to anxiety rather than to depression.

One may argue that the negative attention and interpretation bias (sometimes) shown by depressed subjects might be mediated by co-occurring enhanced anxiety levels, that is, a negative bias is related with anxiety rather than with depression. Following this reasoning, variation in anxiety levels may explain two phenomena. First, it may account for the discrepancies found between studies on biases in perception of facial expressions. Evidence on negative biases in perception of facial expressions in depression is equivocal. Some authors found a depression related negative bias and others did not (respectively Mandal and Palchoudhury, 1985, Mandal and Bhattacharya, 1985, Gur et al., 1992, Rubinow and Post, 1992versus Walker et al., 1984, Zuroff and Colussy, 1986, Mandal, 1987, Archer et al., 1992, Gaebel and Woelwer, 1992, Bouhuys et al., 1996). Second, negative biases of high anxious depressed subjects may explain the often reported unfavourable course of depression in these subjects: The negative biases in the judgment of nonverbal behaviour may affect interpersonal processes negatively, resulting in persistence of depression.

Most research on cognition and emotion has been done on non-clinical populations, leaving unanswered the question as to whether in a clinically depressed population co-occurring anxiety may explain changes in perception. Moreover, as far as we know, the underlying assumption that anxiety-related cognitions may play a role in depression persistence has never been tested in a clinical population. The present study will go in more detail on these issues. We will investigate facial perception of negative and positive emotions in subjects with a primary depression differing in co-occurring anxiety levels. Moreover, we will investigate whether baseline perception of facial emotions is related with the subsequent course of depression, which may reveal some light on factors that determine depression course. In a longitudinal design, depression, anxiety and the perception of facial expressions will be assessed at admission (T0), and after 6 weeks (T1) and 30 weeks (T2) later.

The following results are anticipated:

• Levels of anxiety in clinically depressed patients are positively related with the perception of negative facial emotions: We anticipate that these relationships exist across subjects (at T0, T1, and T2) as well as within subjects (covariation of perception when anxiety and/or depression changes); No relationships between anxiety and the processing of positive facial expressions are expected.

• This negative bias is more strongly associated with levels of anxiety than with levels of depression.

• The putative anxiety-related negative bias predicts an unfavourable course of depression. If so, then it is more likely that differences in selective processing represent the cognitive substrate of vulnerability to emotional disorder persistence.