نشانه توجه بصری با حالات احساسی چهره: تاثیر تفاوت های فردی در اضطراب و افسردگی

Abstract Theoretical models on the development of anxiety disorders and depression have postulated mood-congruent information processing biases as a cognitive vulnerability factor. Hierarchical models of these disorders suggest shared and distinct cognitive processing biases in at-risk individuals. In the present study, attentive processing of emotional facial expressions was investigated in a large group of participants (N = 144) that were tested on tension/stress, anxiety, and depression symptoms. In a modified version of the exogenous cueing paradigm, spatial attention was cued by an angry, sad, happy, or neutral facial expression that correctly or incorrectly predicted the location of a target. Results showed no main or interaction effects of emotional expression and individual differences on attentional cueing. The absence of any attentional cueing effects is discussed in relation to population characteristics and previous null-results in the attentional bias literature.

. Introduction An important theme in research on emotional disorders is the distinction between normal and biased orienting of attention towards emotional information. Given the importance of emotional information in guiding our actions, it has been argued that emotionally-laden stimuli demand attention in everyone (e.g., Eccleston and Crombez, 1999 and Mogg and Bradley, 1998). This claim holds in particular for threatening information which requires fast attentional orienting to the source of danger in order to maximize the chances of successful responding (Lang et al., 1997 and Öhman et al., 2000). In patients with emotional disorders, this tendency to orient attention towards emotional information is often enhanced. It has been robustly demonstrated that anxiety patients have an attentional bias in favour of threat-related information (see Mogg and Bradley, 1998 and Williams et al., 1997). In depressed individuals the empirical data on attentional bias is less robust. However, there are indications for an attentional bias in favour of negative information and reduced attention for positive information at later stages of information processing (see Suslow & Dannlowski, 2005). Cognitive theories have proposed that biased attention for emotional information plays an important role in the aetiology of clinical anxiety and depression (Beck, 1967, Beck, 1976 and Williams et al., 1997). More specifically, individuals with a heightened predisposition for anxiety and depression would also be characterized by attentional biases which might contribute to an enhanced emotional reactivity to stress. Research on attentional bias and anxiety vulnerability has mostly focused on high trait anxious individuals. Trait anxiety is a personality factor that predisposes anxious responding to stressful and novel situations ( Spielberger, 1966). Trait anxiety is generally measured by the State and Trait anxiety Inventory (STAI; Spielberger, Gorsuch, Lushene, Vagg, & Jacobs, 1983). Research on attentional bias in depression vulnerability has mostly examined dysphoric individuals. These individuals are often selected on the basis of their scores on the Beck Depression Inventory (BDI; Beck, Ward, Mendelson, Mock, & Erbaugh, 1961). Importantly, research on attentional bias in anxiety- and depression-vulnerability has thus far been largely independent from each other. This is problematic for two main reasons. First, studies in personality constructs underlying anxiety and depression have acknowledged the high co-morbidity between anxiety and depression by proposing that similar personality constructs underlie anxiety and depression. In the influential tripartite model, Clark and Watson (1991) proposed that negative affect (or general distress) underlies both anxiety and depression, with depression being additionally related to lack of positive affect or anhedonia and anxiety being related to somatic hyperarousal. Other models have further refined the taxonomy of personality factors unique for each disorder, but also incorporate the idea of shared general distress in anxiety and depression (e.g., Brown et al., 1998 and Zinbarg and Barlow, 1996). A second, related problem of separate research into anxiety and depression, involves the high correlation between measures for anxiety and depression (see Beuke, Fischer, & McDowall, 2003). Therefore, in many studies on attentional bias using for instance the STAI or BDI it is unclear whether results can be attributed to the contribution of anxiety or depression. Furthermore, it has been noted that heightened anxiety or depression scores may mask the effects of the other construct. For instance, in clinical studies on depression the inclusion or exclusion of individuals with heightened anxiety strongly influenced attentional bias (Musa, Lépine, Clark, Mansell, & Ehlers, 2003). Hence, measures are required that allow the examination of separate effects of anxiety and depression on attentional bias. In the current study, we aimed to elucidate the effects of anxiety, depression, and distress on attentional cueing by emotional facial expressions. Seen from an evolutionary perspective, emotional facial expressions are considered powerful social stimuli with a high capability of influencing the orienting of attention (e.g., Fox et al., 2000 and Vuilleumier and Schwartz, 2001). These stimuli allow examination of attentional effects for both anxiety- and depression-related information. That is, previous research suggests that anxiety and depression may be characterized by mood-congruent attentional biases. In previous research, anxiety has been associated with an attentional bias for threatening information such as angry faces (e.g., Fox, Russo, Bowles, & Dutton, 2001), whereas attentional bias in depression seems specific for negative and positive information, for instance sad and happy facial expressions (e.g., Gotlib et al., 2004, Gotlib et al., 2004 and Suslow and Dannlowski, 2005). In order to investigate the proposed attentional effects, four categories of emotional facial expressions – angry, sad, happy, and neutral faces – were presented in an exogenous cueing paradigm (Posner, 1980). In this task, participants are asked to detect a visual target presented at the left or right side of a fixation cross. On most of the trials, a peripheral cue precedes the target at the same spatial location (“valid” trials). On the remaining trials, the target is presented at the opposite spatial location of the cue (“invalid” trials). Exogenous cues that are presented for a short duration facilitate responding to target stimuli on valid trials, whereas on invalid trials a reaction time cost is observed. This pattern is referred to as the “cue validity effect”. In the emotional modification of this paradigm, the emotional value of the cue is varied (i.e., angry, sad, happy, neutral) which allows the investigation of facilitated cueing of attention (‘attentional engagement”) by emotional cues compared with neutral cues. Enhanced attentional engagement by emotion is reflected by reaction time benefits in responding to valid emotional trials compared with valid neutral trials. In addition, this task allows examination of attentional disengagement from emotional cues compared with neutral cues by comparing reaction time costs in responding to invalid emotional trials compared with invalid neutral trials. The modified cueing paradigm has been applied successfully in several studies investigating anxiety- and depression-vulnerability. Studies including trait anxiety found that high anxious individuals show enhanced attentional engagement with and impaired disengagement from threat (Fox et al., 2002, Koster et al., in press and Yiend and Mathews, 2001). In dysphoria, a cueing study using negative, positive and neutral self-referring words indicated that dysphorics maintained attention at negative words on long (1500 ms) but not on short presentation durations. These results are probably related to a difficulty to disengage attention from negative material. Moreover, at the longer presentation duration, researchers also found reduced attention for positive words (Koster, De Raedt, Goeleven, Franck, & Crombez, 2005). Altogether, the empirical research on mood-congruent attentional bias suggests a different pattern in anxiety compared with depression with regard to the time-course of information processing. Therefore, in this study facial expressions were presented for a short (200 ms) and longer (1000 ms) duration. In order to measure individual differences in anxiety and depression the Depression and Anxiety Symptom Scale (DASS; Lovibond & Lovibond, 1995) was used in the current study. Resembling the structure of tripartite and hierarchical models of anxiety and depression, this scale provides a measure of stress, anxiety, and depression symptoms. Notably, the anxiety and depression scales of the DASS were designed to discriminate between these two concepts and in comparison to the BDI and Beck Anxiety Inventory (BAI; Beck & Steer, 1990) these scales showed a greater separation in factor loadings (Lovibond & Lovibond, 1995). The DASS has been validated in undergraduate and clinical samples and has a good reliability, convergent and discriminant validity (Lovibond & Lovibond, 1995). In this study, the DASS and the attentional task are administered in a large undergraduate sample. In this way it is possible to examine whether attentional bias for emotional facial expressions vary dimensionally with anxiety, depression, or stress scores. Additionally, we investigated the attentional effects in the extreme groups of each scale. On the basis of previous research using the modified cueing task, we made the following predictions: • High anxiety would be related to enhanced engagement with and a difficulty to disengage from angry facial expressions presented for 200 ms. • High depression would be associated with maintained attention to and a difficulty to disengage from sad facial expressions presented for 1000 ms. Furthermore, in this group reduced attention for positive facial expressions was expected. Given that only very few studies investigated attentional bias in the function of stress we had no specific hypothesis for the effects of stress on attentional bias.

3. Results 3.1. Data preparation Because the experiment was run in groups of 20–40 participants, we carefully examined erroneous and outlying responses. Criteria for exclusion were >10% erroneous responses, or >15% outlying responses. Overall, few errors were made, M = 0.82%. Two participants made too many errors and were excluded. Outlying responses were response latencies <200 ms or >1000 ms indicating anticipatory and delayed responding, respectively. Mean number of outliers was 2.96% and another three participants were excluded on the basis of >15% outlying responses. Thus, analyses were performed on 144 participants. In this group of participants, responding to the digit trials was accurate (7.23% errors). 3.2. Questionnaire data The raw DASS scores for the included participants are presented in Table 1. Using the cut-off score guidelines (Lovibond & Lovibond, 1995), percentile scores were used to classify participants as normal (0–78 percentiles), mild (78–87 percentiles), moderate (87–95 percentiles), severe (95–98 percentiles), and extremely severe (98–100 percentiles) symptomatic on each subscale (see Table 1). Table 1. Summary statistics for the DASS M SD Range Participants in each DASS category Normal Mild Moderate Severe Extremely severe Total sample (N = 144) Depressive symptoms 5.31 6.22 0–33 119 11 7 4 3 Anxiety symptoms 3.81 4.17 0–19 123 9 6 6 0 Co-morbid symptoms 9.04 6.80 0–31 119 11 9 5 0 Table options 3.3. Response latencies An initial ANOVA was performed to examine any main or interaction effects of picture valence, presentation duration and cue validity on response latencies, unrelated to anxiety and depression. This analysis revealed only a main effect of presentation duration, F(1, 142) = 111.75, p < .001, η2 = .44, with faster responding on the 1000 ms picture presentation (M = 376 ms) compared with the 200 ms picture presentation (M = 405 ms). No other effects were significant (Fs < 1.4). 3.4. Correlational analyses Three indices were calculated for each emotional expression to examine the attentional effects: (1) Cue validity effects [=RTinvalid cue − RTvalid cue] for emotional expressions were compared with the cue validity effects for neutral expressions. (2) Emotional modulation of attentional engagement [=RTvalid/neutral cue − RTvalid/emotional cue]. (3) Emotional modulation of attentional disengagement [=RTinvalid/emotional cue − RTinvalid/neutral cue]. On all three indices, a positive score indicates that attention is more strongly cued by the emotional cue compared with the neutral cue. These indices were calculated separately for each class of emotional expressions and presentation duration and are used throughout the remainder of the analyses. As expected, anxiety and depression scores were significantly correlated, r = .52, p < .001. Anxiety and depression were also correlated to the stress score, r = .63, p < .001, r = 66, p = .001, respectively. Contrary to the hypotheses, correlational analysis revealed no significant effects between attention for emotional expressions and anxiety, depression, stress or DASS total scores (all df’s 143, p’s > .10). All correlations were smaller than .2. 3.5. Extreme group analysis The absence of significant correlations may be attributed to the large group of individuals with “normal” scores on anxiety, depression, and stress (see Table 1). Therefore, we examined whether any individual differences in attentional cueing by emotional expressions could be observed in the individuals with extreme high and low scores on anxiety, depression and stress. Cue validity, engagement, and disengagement indices of individuals scoring “mild” to “extremely severe” were compared with individuals scoring lowest on each subscale. Simple t-tests indicate that the questionnaire scores of high and low groups differed significantly for depression (Mhigh = 17, Mlow = 0), t(48) = 42.85, p < .001; anxiety (Mhigh = 12, Mlow = 0), t(40) = 68.65, p < .001; stress (Mhigh = 21, Mlow = 2), t(48) = 32.82, p < .001. Separate ANOVA’s were performed for each group on all three indices. These ANOVA’s revealed no main or interaction effects involving anxiety, depression or stress (all Fs < 1.8)