اثر آموزش توجه بر کورتیزول و واکنش آلفا آمیلاز بزاقی به استرس روانی: اهمیت کنترل توجه

Summary Introduction This study examined the effects of three consecutive days of attentional training on the salivary alpha amylase (sAA), cortisol, and mood response to the Trier Social Stress Test (TSST). The training was designed to elicit faster disengagement of attention away from threatening facial expressions and faster shifts of attention toward positive ones. Method Fifty-six healthy participants between the ages of 18 and 30 participated in a double-blind, within-subject experiment. Participants were randomly assigned to one of three attentional training conditions – supraliminal training: pictures shown with full conscious awareness, masked training: stimuli presented with limited conscious awareness, or control training: both supraliminal and masked pictures shown but no shifting of attention required. Following training, participants underwent the TSST. Self-reported mood and saliva samples were collected for the determination of emotional reactivity, cortisol, and sAA in response to stress post-training. Results Unexpectedly, participants in both attentional training groups exhibited a higher salivary cortisol response to the TSST relative to participants who underwent the control training, F (4, 86) = 4.07, p = .005, View the MathML sourceηp2=.16. Supraliminal training was also associated with enhanced sAA reactivity, F (2, 44) = 13.90, p = .000, View the MathML sourceηp2=.38, and a more hostile mood response (p = .021), to the TSST. Interestingly, the effect of attention training on the cortisol response to stress was more robust in those with high attentional control than those with low attentional control (β = −0.134; t = −2.24, p = .03). Conclusion This is among the first experimental manipulations to demonstrate that attentional training can elicit a paradoxical increase in three different markers of stress reactivity. These findings suggest that attentional training, in certain individuals, can have iatrogenic effects

. Introduction It is well known that anxiety disorders and other mental illnesses such as depression are associated with abnormalities in the selective processing of emotional stimuli (MacLeod et al., 1986, Williams et al., 1996, Mathews and Macleod, 2005 and Peckham et al., 2010). It is not known, however, whether alterations in emotional information processing represent a correlate or symptom of the disordered state, or contribute to the development or maintenance of these disorders. Recently, randomized-control studies attempting to experimentally modify the focus of attention, known as attentional training, have been conducted in an effort to clarify the issue of causality. If the relationship between attentional abnormalities and anxiety is etiological, then changes in attention should elicit a decrease in anxiety and symptom relief. Indeed, attentional training has been effective in reducing symptoms of anxiety ( Bar-Haim, 2010, Li et al., 2008 and Hakamata et al., 2010), as well as emotional reactivity to stress ( MacLeod et al., 2002 and Amir et al., 2008). Less is known, about the link between attentional biases, attentional training, and physiological markers of stress, as well as the role of attentional control in these associations. Several studies have shown that baseline, early stage attentional biases predict cortisol responses to stress (Ellenbogen et al., 2006, Fox et al., 2010, Roelofs et al., 2007 and Pilgrim et al., 2010). Fox et al. (2010) showed that baseline attentional biases predict cortisol reactivity eight months later while controlling for neuroticism, trait anxiety, and depression. Similarly, faster shifts of attention toward masked angry faces during stress, a marker of automatic processing, was associated with higher cortisol in response to the Trier Social Stress Test (TSST; Ellenbogen et al., 2010). These findings suggest that attention training, particularly at early stages, may influence physiological indices of stress in important ways. Although techniques such as cognitive behavioral stress management (Gaab et al., 2003 and Gaab et al., 2006) and meditation (Lutz et al., 2008 and Sood and Jones, 2013) may also attenuate cortisol levels (Brand et al., 2012, Fan et al., 2014, Gaab et al., 2003, Gaab et al., 2006 and Regehr et al., 2013) and heart rate (Nyklíček et al., 2013), attentional training, which specifically aims to change attentional biases, may shed light on the causality of the relationship between attention and stress. Few studies have examined the impact of attentional training on physiological markers of stress (Dandeneau et al., 2007, Baert et al., 2012, Heeren et al., 2012 and Higgins and Hughes, 2012). Training toward happy faces increased self-esteem and lowered basal cortisol in young adults working in a stressful environment (Dandeneau et al., 2007), while six days of training facilitated disengagement away from threat and improved heart rate variability (a marker of autonomic system functioning) during the recovery phase of a simulated job interview (Baert et al., 2012). Despite these results the exact mechanism by which attentional training occurs, remains largely unknown (Heeren et al., 2013). Attentional biases may be the result of a valence-specific system which modifies initial threat detection (Heeren et al., 2013) and/or sharpens attentional control (Heeren et al., 2013, Klumpp and Amir, 2010 and Paulewicz et al., 2012) which, rather than reducing early threat processing would modify responses to incoming threat. According to this view, attentional control regulates bottom-up emotional responses. This is supported by evidence that attentional control moderates the link between trait anxiety and fear during exposures to biological challenge (i.e., a single inhalation of 35% CO2 enriched gas; Richey et al., 2012). Derryberry and Reed (2002) also showed that strong attentional control predicts faster disengagement from threat and/or engagement toward non-threatening information, irrespective of trait anxiety. Similarly, the strength of the relationship between trait anxiety and amygdala activation in response to stress was moderated by attentional control (Bishop, 2009). These findings suggest that attentional control may play a role in the top-down regulation of stress. Taken together, there is some evidence that attentional training influences stress reactivity, but more research is necessary. Given the link between rapid threat processing and cortisol reactivity observed across several studies (Ellenbogen et al., 2006, Ellenbogen et al., 2010 and Fox et al., 2010), it is possible that training early attentional processes may influence stress reactivity in unique ways. The present study compared training of early versus later stages of attention on the stress response as indexed by the subjective emotional response and two major biological stress systems – the sympathetic nervous system (SNS) and the hypothalamic–pituitary–adrenal (HPA) axis. The SNS is triggered immediately following stress, representing an early stage of the stress response, while the HPA axis, which responds 20–30 min later (Chrousos, 2009), reflects a later stage of reactivity or adaptation to an ongoing challenge. Both salivary alpha amylase (sAA), a digestive enzyme and marker of SNS activity (Nater and Rohleder, 2009), and salivary cortisol, the major human glucocorticoid stress hormone secreted by the HPA axis, will be collected in the present study. It is important to determine if attentional training has specific effects on the HPA axis, consistent with some past correlational research (Ellenbogen et al., 2006), or whether it influences stress reactivity across all systems (mood state, cortisol, and sAA). In the present study, healthy participants were randomly assigned to one of three attention training manipulations: a masked condition using briefly presented stimuli (17 ms) followed by a mask (733 ms; limited conscious awareness), a supraliminal condition using pictures displayed for 750 ms (full conscious awareness), and a control condition where participants were exposed to both picture types but not required to shift attention. Attentional training took place over three days and was aimed at facilitating attentional disengagement away from threat (angry faces) and promoting shifts of attention toward positive stimuli (happy faces). The hypotheses were: (1) attention training will elicit faster disengagement from threat, (2) attention training will attenuate mood, cortisol, and sAA reactivity post-training, (3) attention training with masked stimuli will have a more robust effect on the cortisol response to stress than supraliminal training, and (4) baseline attention control will moderate the association between attentional training and parameters of stress reactivity.

3. Results 3.1. Baseline measures One-way ANOVAs revealed that the training groups differed on baseline depression on the Beck Depression Inventory_II, F (2, 52) = 4.23, p = .02. Tukey's HSD tests revealed that the supraliminal group had significantly higher scores on the Beck Depression Inventory_II relative to the masked training, p = .038, and control conditions, p = .038. Therefore, we covaried for the Beck Depression Inventory_II in all subsequent analyses. 3.2. Impact of attentional training on selective attention A Group × Sex MANCOVA was conducted on the six engagement difference scores (Day 5 − Day 1) for trials with angry and sad faces presented for 17 ms (masked), 200 ms and 750 ms durations. The multivariate test for group was not significant but the between-subject univariate ANCOVAs revealed group differences for masked stimuli depicting threat presented for 17 ms, F (2, 38) = 4.70, p = .015; View the MathML sourceηp2=.20. There were no differences for stimuli depicting threat presented for 200 ms or 750 ms, nor for trials with sad faces. Planned simple contrasts revealed that both the supraliminal (p = .011, CI [−.055, −.007]) and the masked training groups (p = .024, CI [−.044, −.003]) were slower to shift attention toward masked angry faces following training relative to the control condition. Bonferroni-corrected pairwise comparisons revealed no differences between the supraliminal and masked training groups on trials with angry faces. A Group × Sex MANCOVA on the six difference scores for disengagement from angry and sad faces presented for 17 ms, 200 ms and 750 ms revealed a Group × Sex interaction for trials with angry faces, F (2, 38) = 3.16, p = .054, View the MathML sourceηp2=.14, presented for 750 ms. These results were followed up by an ANCOVA on trials with stimuli depicting threat presented for 750 ms in male and female participants. Results revealed a significant sex difference only in the supraliminal group, F (1, 16) = 12.08, p = .007, View the MathML sourceηp2=.57. Simple contrasts revealed that females were faster to disengage from angry faces following supraliminal training than males (p = .007, CI [−.089, −.019]). In sum, the supraliminal and masked training conditions were successful in reducing attentional engagement toward masked angry faces. Supraliminal attention training also elicited more efficient disengagement from angry faces, presented with full conscious awareness (750 ms), in females compared to male participants. There were no training related effects or interactions on trials with sad faces. 3.3. Attentional training and the salivary cortisol response to stress A Group × Sex × Time mixed design ANCOVA on cortisol levels revealed a main effect of time, F (2, 86) = 24.45, p = .000, View the MathML sourceηp2=.36, and a Group × Time interaction, F (4, 86) = 4.07, p = .005, View the MathML sourceηp2=.16 ( Fig. 2). The interaction was followed up with a Group × Sex ANCOVA for AUCi cortisol, which revealed a significant main effect of group, F (2, 44) = 5.07, p = .010, View the MathML sourceηp2=.19. Cortisol AUCi (mean ± SD) was 11.74 ± 8.33, 11.04 ± 9.78, and 5.55 ± 4.53, for the supraliminal, masked and control training groups, respectively. Planned simple contrasts showed that both the supraliminal (p = .008, CI [2.07, 12.81]) and the masked training groups (p = .015, CI [1.33, 11.42]) had higher AUCi cortisol than participants who were in the control condition. Bonferroni-corrected pairwise comparisons revealed no additional differences between supraliminal versus masked training. Salivary cortisol reactivity to the TSST across the three groups (i.e., ... Fig. 2. Salivary cortisol reactivity to the TSST across the three groups (i.e., supraliminal training, masked training and control). Bars represent standard error of the mean. Depicts values obtained before covarying for the Beck Depression Inventory_II. Figure options Analyses were conducted on AUCg, but no group differences were found (data not shown). Cortisol AUCg (mean ± SD) was 20.62 ± 8.95, 21.15 ± 14.91, and 18.08 ± 16.55, for the supraliminal, masked and control training groups, respectively. Additional analyses were conducted to examine whether oral contraceptive use may have altered the findings reported above. The Group × Sex × Time analyses on cortisol levels were repeated covarying for oral contraceptive use: both the main effect, F (2, 94) = 34.66, p = .000, View the MathML sourceηp2=.42, and Group × Time interaction, F (4, 94) = 4.09, p = .004, View the MathML sourceηp2=.15, were retained. In summary, the TSST elicited a significant increase in cortisol from baseline. Attentional training in both experimental groups increased the magnitude of the cortisol response to the TSST relative to participants in the control condition. 3.4. Attentional training and the salivary alpha amylase response to stress The Group × Sex × Time mixed-design ANCOVA on sAA levels revealed a main effect of time, F (4, 161) = 8.08, p = .000, and group, F (2, 44) = 3.98, p = .026, partial View the MathML sourceηp2=.15 ( Fig. 3), but no significant interactions were found. The main effect of group was followed up with a Group × Sex ANCOVA for AUCi sAA, which revealed a significant main effect of training group, F (2, 44) = 13.90, p = .000, View the MathML sourceηp2=.38. sAA AUCi (mean ± SD) was 10527 ± 5898, 5378 ± 4060, and 4253 ± 3818, for the supraliminal, masked and control training groups, respectively. Planned simple comparisons revealed that the supraliminal training group had higher levels of sAA than the control group (p = .000, CI [4974, 11,254]), but there were no differences between the masked training group and controls. Bonferroni-corrected pairwise comparisons revealed that the supraliminal group also displayed higher AUCi sAA than the masked training group (p = .002). Salivary alpha amylase to the TSST across the three groups (i.e., supraliminal ... Fig. 3. Salivary alpha amylase to the TSST across the three groups (i.e., supraliminal training, masked training, and control). Bars represent standard error of the mean. Depicts values obtained before covarying for the Beck Depression Inventory_II. Figure options A Group × Sex ANCOVA on AUCg sAA also revealed a significant main effect of group, F (2, 44) = 7.81, p = .001, View the MathML sourceηp2=.26. sAA AUCg (mean ± SD) was 20,261 ± 10,244, 11,373 ± 7767, and 10,560 ± 6122, for the supraliminal, masked and control training groups, respectively. Planned simple comparisons revealed that the supraliminal training group had higher AUCg sAA than the control group (p = .000, CI [5516, 17,675]), but there were no differences between the masked training and control groups. Bonferroni-corrected pairwise comparisons revealed that the supraliminal condition had elevated AUCg sAA relative to the masked group (p = .013). There were no significant sex differences. In summary, attentional training with supraliminal stimuli, but not with masked stimuli, elicited an increased sAA response to the TSST relative to the control condition. 3.5. Attentional training and mood response to stress The Group × Sex MANCOVA on POMS difference scores, did not reveal a significant main effect of group, but univariate tests showed a marginally significant group difference for hostility, F (2, 45) = 2.89, p = .07, View the MathML sourceηp2=.11. Planned simple contrasts showed elevated hostility post-TSST in the supraliminal training group relative to the control group (p = .021, CI [−8.26, −.722]). There were no significant sex differences. In sum, participants who underwent attention training exhibited marginally more hostility in response to the TSST than control participants. 3.6. Regression: baseline attentional control and stress reactivity In order to examine whether baseline attentional control moderates the observed association between attentional training and cortisol reactivity, a hierarchical multiple regression was performed using cortisol AUCi as the outcome. To reduce the number of predictor variables in the equation, the training groups were collapsed into a single group (coded as follows: any training = 1; control = 0). In the first step, centered baseline mean attentional control scores, Beck Depression Inventory_II scores, and training group were entered. In the second step, the training group by baseline attentional control score interaction was added. Lastly, in the third step the training group by Beck Depression Inventory_II Score interaction was added. The regression equation was significant [n = 44; R = .599, F (5, 44) = 4.91, p = .001], accounting for 28% (adjusted R2) of the variance. Both attentional training group (β = 7.44; t = 3.45, p = .01) and the interaction between baseline attentional control and attentional training group (β = −0.134; t = −2.24, p = .03) were significant predictors of the AUCi. Thus, baseline attentional control moderated the relationship between attention training and cortisol change in response to stress ( Table 2). Table 2. Standardized regression coefficients (B) for baseline attentional control and the interaction between group and baseline attentional control in the prediction of cortisol area under the curve with respect to increase (controlling for baseline Beck Depression Inventory_II Score). B SE B β Step 1 Constant 4.615 1.651 Attentional control −.065 .03 −.299* Beck Depression Inventory_II −.164 .152 −0.146 Attentional training versus control 7.852 2.134 .473* Step 2 Constant 5.68 1.67 Attentional control .005 .045 .024 Beck Depression Inventory_II −.103 .15 −.092 Attentional training versus control 7.038 2.096 .424 Attentional training × Baseline Attentional Control −.119 .057 −.423* Step 3 Constant 5.312 1.733 Attentional control .013 .046 .06 Beck Depression Inventory_II −.366 .348 −.327 Attentional training versus control 7.436 2.156 .448 Attentional training × Baseline Attentional Control −.134 .06 −.474* Attentional training × Beck Depression Inventory_II .324 .386 .263 Table options Simple slope analyses (Aiken and West, 1991) were performed to follow up the significant attentional control × attention training interaction. The slope for participants with high attentional control (one standard deviation below the mean) was significantly different than zero, t(48) = 4.21, p = .000 (−1 SD; β = .77), indicating that attention training increased cortisol reactivity in those with stronger attentional control relative to the control group ( Fig. 4). The slope for participants with low attentional control scores (one standard deviation above the mean), in contrast, did not differ significantly from zero (+1 SD; β = .16), indicating that those with low attentional control did not exhibit any training-related changes in cortisol. No moderation effects were found for sAA (data not shown). Simple slopes predicting the cortisol AUCi response to the TSST post-training ... Fig. 4. Simple slopes predicting the cortisol AUCi response to the TSST post-training (i.e., training groups combined) versus control for 1 SD below the mean for attentional control and 1 SD above the mean for attentional control.