اصلاح انحراف توجه تسهیل کننده مکانیسم کنترل توجه: شواهدی از ردیابی چشم
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|38695||2015||8 صفحه PDF||سفارش دهید||7222 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Biological Psychology, Volume 104, January 2015, Pages 139–146
Abstract Social anxiety is thought to be maintained by biased attentional processing towards threatening information. Research has further shown that the experimental attenuation of this bias, through the implementation of attentional bias modification (ABM), may serve to reduce social anxiety vulnerability. However, the mechanisms underlying ABM remain unclear. The present study examined whether inhibitory attentional control was associated with ABM. A non-clinical sample of participants was randomly assigned to receive either ABM or a placebo task. To assess pre–post changes in attentional control, participants were additionally administered an emotional antisaccade task. ABM participants exhibited a subsequent shift in attentional bias away from threat as expected. ABM participants further showed a subsequent decrease in antisaccade cost, indicating a general facilitation of inhibitory attentional control. Mediational analysis revealed that the shift in attentional bias following ABM was independent to the change in attentional control. The findings suggest that the mechanisms of ABM are multifaceted.
. Introduction Social anxiety disorder (SAD) is a debilitating form of mental illness characterized by an excessive fear of negative social evaluation. Cognitive theories have emphasized the role of aberrant information processing in the maintenance and exacerbation of this condition, with particular emphasis on early attentional processing (Mogg and Bradley, 1998 and Williams et al., 1997). A strong base of empirical literature suggests that SAD is characterized by an attentional bias favouring the processing of social threat stimuli (Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, & van Ijzendoorn, 2007). That is, socially anxious individuals may preferentially allocate attentional resources to the detection of cues in their environment which may indicate social disapproval.1 This attentional bias to threat serves to maintain anxiety by disproportionately elevating arousal in social situations, falsely confirming various negative cognitions, and triggering maladaptive safety behaviours (Rapee & Heimberg, 1997). A causal relationship between attentional bias and anxiety vulnerability has further been demonstrated through the use of attentional bias modification (ABM; MacLeod, Rutherford, Campbell, Ebsworthy, & Holker, 2002). The most commonly used ABM task involves the brief presentation of a threat and neutral stimulus pair, followed by a probe which appears in the location vacated by one of the two stimuli. Participants are required to respond to the probe. However, the probe location is contingent on the stimulus pair such that the probe is always presented in the location of the neutral stimulus. Efficient completion of this task therefore will be enhanced by reducing attentional bias towards threat. MacLeod et al. (2002) demonstrated that this task can indeed modulate attentional bias. This induction of a bias away from threat was further associated with reduced anxious reactivity in response to a subsequent stressor. Recent studies have additionally shown that ABM may be protective against anxiety during a naturalistic stressor (See, MacLeod, & Bridle, 2009), and may reduce social anxiety symptoms in clinically socially anxious individuals (Amir et al., 2009 and Schmidt et al., 2009). Taken together, recent research suggests a causal relationship between ABM and social anxiety vulnerability. While such findings may be suggestive of the potential therapeutic application of ABM in the treatment of SAD, the mechanisms which underlie the modulation of attentional bias are not well understood. Neurocognitive models of attention suggest that two biasing signals, from a stimulus driven system and an attentional control system, determine selective attention to emotional stimuli (Corbetta & Shulman, 2002). The stimulus driven system is largely amygdala-centred, and encompasses bottom-up processing. This system functions to automatically deploy attention to salient stimuli, with particular regard for the detection of threat (Bishop, 2007). The latter attentional control system recruits regions such as the prefrontal cortex (PFC), and may subsequently provide a more flexible top-down control of attention relevant to current goals and task requirements, including for instance, the inhibition of task irrelevant information (Bishop, 2007, Bishop, 2008, Desimone and Duncan, 1995 and Vuilleumier, 2005). It has been suggested that the anxiety-linked attentional bias to threat may represent an imbalance between the two attentional systems. Attentional Control Theory (Eysenck, Derakshan, Santos, & Calvo, 2007) posits that heightened anxiety is associated with an enhanced stimulus driven system and a relatively impaired attentional control system. Correspondingly, anxiety has been associated with amygdala hyperactivation in response to threatening stimuli (Bishop et al., 2004b, Phan et al., 2006, Stein et al., 2007, Straube et al., 2005 and Yoon et al., 2007). Evidence further suggests that anxiety is associated with the impoverished recruitment of prefrontal areas suggesting a deficit in attentional control (Bishop, 2009 and Bishop et al., 2004a). Taken together, anxious individuals may have difficulty exerting goal directed attention in the presence of task irrelevant stimuli. With regard to threat stimuli, such information may be particularly difficult to inhibit (Eysenck et al., 2007). Within this framework, it is possible that ABM may modulate selective attention by affecting either the stimulus driven or attentional control systems. Recent imaging research has observed altered lateral PFC activation in response to the presentation of threat stimuli following ABM training away from threat (Browning, Holmes, Murphy, Goodwin, & Harmer, 2010). Additional electrophysiological research has found that ABM may modulate the event-related potential (ERP) complex (Eldar & Bar-Haim, 2010). In this study, participants who were administered ABM exhibited a subsequent increase in N2 amplitude and reductions in P2 and P3 amplitudes, while no change was observed for early P1 and N1 ERP components. It was inferred that ABM is associated with modulation of later information processing, including increased attentional control, and a reduction in the neurocognitive resources allocated to the processing of the emotional properties of stimuli. Further ERP studies have similarly suggested that ABM may influence later stages of information processing (Osinsky et al., 2014 and Suway et al., 2013), although the modulation of earlier attentional processing following ABM has also been observed (O’Toole & Dennis, 2012). While recent findings generally implicate that ABM may facilitate attentional control, no direct assessment of changes in attentional control performance following ABM has been conducted. However, such potential ABM-related changes in attentional control may be examined by changes in antisaccade task (Hallett, 1978) performance. The antisaccade task is a well established method for the assessment of inhibitory attentional control, and has been previously applied to range of psychopathology (Hutton & Ettinger, 2006). This task involves the presentation of a peripheral stimulus. Participants are required to either prosaccade (look towards) or antisaccade (look away) from the stimulus. While the prosaccade is largely a reflexive response, central to the execution of the antisaccade is the inhibition of the prepotent prosaccade response, prior to the subsequent generation of a volitional saccade (Munoz & Everling, 2004). Critically, correct antisaccade performance necessitates the recruitment of attentional control-related brain regions, such as the lateral PFC, thus supporting its use as an index of inhibitory attentional control (Ettinger et al., 2008, Hutton and Ettinger, 2006 and Munoz and Everling, 2004). Saccade latency and error rates are typically measured. Indeed, the antisaccade task has been a useful tool in the elucidation of the attentional anomaly marked in anxiety. Attentional Control Theory (Eysenck et al., 2007) further posits that while anxiety is characterized by a deficit in attentional control, such a deficit may primarily impair the efficiency domain of processing, which is typically indexed by speed or reaction time measures. In contrast, the effectiveness of processing, inferred from measures such as error rates, is considered to be relatively unimpaired. Consistent with this notion, high trait anxious individuals, in comparison to low anxious individuals, have typically been found to exhibit longer antisaccade latencies, while no differences in error rates were observed ( Ansari and Derakshan, 2009, Ansari and Derakshan, 2011 and Ansari et al., 2008). Such findings suggest that anxiety may be associated with a deficit in the efficiency of attentional control. In order to further examine whether attentional control deficits may be general or more specific with regard to the inhibition of threatening information, an emotional variant of the antisaccade task has previously been employed (Derakshan, Ansari, Hansard, Shoker, & Eysenck, 2009). While the standard antisaccade task uses solely a neutral stimulus, the emotional antisaccade task additionally presents socially relevant emotionally valenced stimuli. Anxious individuals have exhibited longer latencies to antisaccade away from threat stimuli (Derakshan et al., 2009 and Reinholdt-Dunne et al., 2012), suggesting a specific impairment in inhibiting the attentional deployment to threatening information. Given the findings in high trait anxious individuals, social anxiety may be associated with a deficit in inhibitory attentional control, and such a deficit may be particularly apparent for the inhibition of threat stimuli. The present study sought to assess whether ABM may modulate inhibitory attentional control in a non-clinical sample, as assessed by performance changes in an emotional antisaccade task administered before and after ABM. It is well established that antisaccade latencies are longer than prosaccade latencies (Munoz & Everling, 2004). This difference is primarily thought to reflect the inhibition of the prepotent saccade needed in order to perform an antisaccade (Olk & Kingstone, 2003), and is typically referred to as the antisaccade cost ( Godijn and Kramer, 2008 and Reinholdt-Dunne et al., 2012). Given recent findings ( Browning et al., 2010 and Eldar and Bar-Haim, 2010), it was hypothesized that ABM, in comparison to a placebo task, would subsequently increase inhibitory attentional control, indicated by a relative reduction in the antisaccade cost. In light of previous antisaccade research suggesting both general (e.g., Ansari & Derakshan, 2009) and threat-specific (e.g., Derakshan et al., 2009) inhibitory impairments in anxiety, two alternate hypotheses were formed. It was first predicted that ABM would result in a general facilitation of inhibitory attentional control, inferred from a relative reduction antisaccade cost across all levels of stimulus valence following ABM. Alternatively, it was predicted that ABM may result in a threat-specific facilitation of inhibitory attentional control, indexed by a reduction in antisaccade cost for only threat stimuli. A secondary aim of the present study concerned the relationship between attentional bias and inhibitory attentional control. ABM has previously been shown to augment attentional bias (e.g., MacLeod et al., 2002) and for the present study, it was hypothesized that ABM may additionally modulate attentional control. However, the relationship between such ABM-related changes remains to be determined. Hence, the present study sought to examine the possibility that the change in attentional control may mediate the effect of ABM in reducing attentional bias to threat.
نتیجه گیری انگلیسی
Results 3.1. Group characteristics A series of one-way analyses of variance (ANOVA) were conducted to confirm that ABM and placebo assigned participants did not differ in terms of emotional disposition or demographic variables. No group differences were evident, largest F(1,41) = 1.36, p = .250, partial η2 = .03. No group differences in age, F(1,41) = .03, p = .871, partial η2 = .001, or gender ratios, χ2(1, N = 43) = .37, p = .541, were observed. Descriptive statistics are provided in Table 1. Table 1. Means, standard deviations and difference tests for ABM and placebo conditions. Questionnaire ABM Placebo F(1,41) M SD M SD SPS 16.13 6.53 19.85 13.60 1.36 SIAS 23.70 11.41 25.70 12.12 0.31 STAI-T 43.30 8.58 42.65 8.33 0.06 DASS depression 9.22 7.20 8.40 7.16 0.14 DASS anxiety 9.65 8.01 9.20 7.30 0.04 DASS stress 13.74 9.19 14.20 5.98 0.04 Age 22.26 6.61 21.95 5.68 0.03 Table options 3.2. Impact of ABM on attentional bias to threat To assess whether the ABM procedure successfully modified selective attention as required, a group (ABM vs. placebo) by time (pre vs. post) mixed-design ANOVA was conducted on bias scores extracted from the dot probe pre- and post-assessments. Importantly, a significant group by time interaction was evident, F(1,41) = 4.20, p = .047, partial η2 = .09, indicating that at post-assessment, ABM participants showed a bias away from threat in comparison to placebo, t(41) = 2.04, p = .048, partial η2 = .09, while groups did not differ at pre-assessment, t(41) = −.87, p = .390, partial η2 = .02. No other effects were observed, largest F(1,41) = .58, p = .449, partial η2 = .01. Descriptive statistics are provided in Table 2. Table 2. Means and standard deviations of bias scores for ABM and placebo conditions. Negative scores indicate a relative bias away from threat. Bias score ABM Placebo M SD M SD Pre-assessment −2.32 17.34 −7.62 22.67 Post-assessment −9.01 21.32 2.78 15.67 Table options 3.3. Impact of ABM on attentional control To assess the prediction that ABM may modulate antisaccade performance, a group (ABM vs. placebo) by time (pre vs. post) by saccade type (pro- vs. antisaccade) by valence (positive vs. neutral vs. threat) mixed-design ANOVA was run on saccade latency data, shown in Table 3. A main effect of saccade was evident, F(1,41) = 321.27, p < .001, partial η2 = .89, indicating that all participants were slower to correctly perform an antisaccade (M = 221.05, SD = 35.34) compared to a prosaccade (M = 132.27, SD = 18.75). A main effect of time was observed, F(1,41) = 9.71, p = .003, partial η2 = .19, suggesting that all participants were slower to perform the appropriate saccade at post assessment (M = 179.84, SD = 23.13), relative to pre assessment (M = 173.48, SD = 25.04). Of critical importance, a three-way group by time by saccade type interaction was evident, F(1,41) = 4.50, p = .040, partial η2 = .10, confirming that ABM served to modify attentional control. Of further importance, stimulus valence was not found to influence saccade latency, largest F(2,82) = 1.38, p = .258, partial η2 = .03, hence, the effect of ABM on attentional control was equally evident regardless of stimulus valence. Table 3. Saccade latency means and standard deviations for ABM and placebo conditions. Variable ABM Placebo Type Valence M SD M SD Pre-assessment Prosaccade Positive 130.97 21.82 127.22 17.02 Neutral 128.52 17.09 128.43 16.57 Threat 129.46 19.30 128.25 21.03 Antisaccade Positive 218.02 42.23 220.45 38.59 Neutral 219.93 43.25 217.28 35.44 Threat 217.36 43.65 215.93 37.69 Post-assessment Prosaccade Positive 139.58 22.94 132.04 18.99 Neutral 137.60 24.78 133.29 21.90 Threat 139.98 23.12 131.94 19.34 Antisaccade Positive 218.36 35.00 226.34 29.65 Neutral 221.63 38.11 227.30 34.88 Threat 220.14 35.34 229.84 38.82 Table options In order to clarify the nature of this three-way interaction, antisaccade cost scores were computed from pre- and post-saccade assessments. Antisaccade cost was defined as the difference between the mean antisaccade latency and mean prosaccade latency, in which a higher score reflects a greater cost and relatively poorer inhibitory attentional control. Since no influence of stimulus valence was evident, antisaccade cost scores were averaged across all valence levels. A group (ABM vs. placebo) by time (pre vs. post) ANOVA was calculated for antisaccade cost scores. A group by time interaction was observed, F(1,41) = 4.50, p = .040, partial η2 = .10, replicating the initial interaction as expected. Paired samples t-tests confirmed that ABM participants exhibited a reduction in antisaccade cost from pre- to post-assessment, t(22) = 2.19, p = .039, partial η2 = .18, while no change in antisaccade cost was observed for placebo participants, t(19) = −1.03, p = .316, partial η2 = .05. This interaction is illustrated in Fig. 1. Antisaccade latency cost. Changes in antisaccade latency cost following ABM or ... Fig. 1. Antisaccade latency cost. Changes in antisaccade latency cost following ABM or placebo conditions. Cost scores for each group are averaged across all valence levels. Figure options To examine whether ABM induced any changes in saccade error rates, a group (ABM vs. placebo) by time (pre vs. post) by saccade type (pro- vs. antisaccade) by valence (positive vs. neutral vs. threat) mixed-design ANOVA was run on saccade error rates. A main effect of saccade type was found, F(1,41) = 41.75, p < .001, partial η2 = .51, indicating that participants had a higher error rate for antisaccades (M = 11.43, SD = 11.46) compared to prosaccades (M = .29, SD = .23). No other effects were significant, largest F(1,41) = 2.28, p = .139, partial η2 = .05. Descriptive statistics for saccade error rates are presented in Table 4. Table 4. Means and standard deviations for saccade error rates for ABM and placebo conditions. Variable ABM Placebo Type Valence M SD M SD Pre-assessment Prosaccade Positive 0.00 0.00 0.00 0.00 Neutral 0.00 0.00 0.00 0.00 Threat 0.00 0.00 0.29 1.32 Antisaccade Positive 7.62 7.80 16.47 13.71 Neutral 9.98 10.30 11.82 14.81 Threat 9.33 9.48 15.75 15.93 Post-assessment Prosaccade Positive 0.00 0.00 0.95 4.26 Neutral 0.00 0.00 1.05 3.80 Threat 0.00 0.00 1.14 5.08 Antisaccade Positive 9.02 10.63 13.42 16.11 Neutral 8.75 12.44 12.76 13.76 Threat 9.33 14.26 12.93 14.18 Table options 3.4. Mediational influence of attentional control on ABM and attentional bias It was further sought to examine whether the ABM-induced change in attentional bias to threat (ABΔ) was mediated by the change in inhibitory attentional control (ACΔ). A bias change score and an antisaccade cost change score were calculated by subtracting the respective pre-assessment score from the respective the post-assessment score. Analysis was initially conducted using the four step approach outlined by Baron and Kenny (1986). The indirect effect was then tested for significance using the bias-corrected bootstrap procedure by Preacher and Hayes (2004). Fig. 2 depicts the model examined. Standardized coefficients have been reported. Mediational model. Mediational model for the change in attentional control as a ... Fig. 2. Mediational model. Mediational model for the change in attentional control as a mediator between ABM and the change in attentional bias. Standardized regression coefficients are reported (*p < .05). The indirect effect (ab) was not found to be significant at the 95% confidence interval. Figure options The procedure described by Baron and Kenny (1986) requires the following criteria to be met for mediation to occur: (1) the independent variable (ABM) significantly predicts the proposed mediator (ACΔ), described as path a, (2) the proposed mediator (ACΔ) significantly predicts the dependent variable (ABΔ) controlling for the independent variable (ABM), described as path b, (3) the independent variable (ABM) significantly predicts the dependent variable (ABΔ), described as path c, and (4) the relationship between the independent variable (ABM) and the dependent variable (ABΔ) becomes significantly minimized when the proposed mediator (ACΔ) is added to the regression model, described as path c′. (1) Path a was tested by regressing ACΔ on ABM. ABM was found to significantly predict ACΔ, β = −.32, t(41) = −2.12, p = .040. (2) Path b was examined by regressing ABΔ on ACΔ while controlling for ABM, and was not found to be significant, β = .08, t(41) = .52, p = .608. (3) Path c was then examined by regressing ABΔ on ABM, and ABM was found to significantly predict ABΔ, β = −.30, t(41) = −2.05, p = .047. (4) Finally, path c’ was examined. ABM was not found to predict ABΔ, while controlling for ACΔ, β = −.28, t(41) = −1.77, p = .085. Given that path b was not significant, mediation was not likely to have occurred. We further directly examined the effect of ABM on ABΔ through ACΔ (i.e. the indirect effect) by using the bootstrapping procedure outlined by Preacher and Hayes (2004), as this method may provide greater power for small sample sizes and require less assumptions compared to the Baron and Kenny (1986) procedure (Preacher & Hayes, 2004). 5000 bootstrap resamples were used. The 95% confidence interval ranged from −.13 to .06. The inclusion of zero within this interval suggests that ACΔ did not mediate the relationship between ABM and ABΔ.