Perceptual load modulates anterior cingulate cortex response to threat distractors in generalized social anxiety disorder

Abstract

Generalized social anxiety disorder (gSAD) is associated with impoverished anterior cingulate cortex (ACC) engagement during attentional control. Attentional Control Theory proposes such deficiencies may be offset when demands on resources are increased to execute goals. To test the hypothesis attentional demands affect ACC response 23 patients with gSAD and 24 matched controls performed an fMRI task involving a target letter in a string of identical targets (low load) or a target letter in a mixed letter string (high load) superimposed on fearful, angry, and neutral face distractors. Regardless of load condition, groups were similar in accuracy and reaction time. Under low load gSAD patients showed deficient rostral ACC recruitment to fearful (vs. neutral) distractors. For high load, increased activation to fearful (vs. neutral) distractors was observed in gSAD suggesting a compensatory function. Results remained after controlling for group differences in depression level. Findings indicate perceptual demand modulates ACC in gSAD.

Introduction

Generalized social anxiety disorder (gSAD) is characterized by pervasive fears of negative evaluation (APA, 2000) and attentional bias toward threat-relevant stimuli (Bögels & Mansell, 2004) making it difficult to ignore such stimuli even in the context of cognitively demanding tasks (Hope et al., 1990, Mattia et al., 1993, Spector et al., 2003). According to biased competition models of attention, sensory-driven emotional signals compete with task-relevant demands for resources in a limited-capacity processing system (Desimone and Duncan, 1995, Pessoa et al., 2002). What prevails, even if incongruent to cognitive aims, subsequently interacts with emotion-generating regions; therefore, prefrontal areas that modulate attentional deployment such as the anterior cingulate cortex (ACC) (Bush et al., 2000, Etkin et al., 2011) play a role in emotion generation and regulation (Ochsner, Silvres, & Buhle, 2012). Attentional bias is a proposed causal mechanism in maintaining anxiety that is excessive (e.g., Mathews and MacLeod, 1994, Williams et al., 1997); consequently, it is important to elucidate mechanisms associated with bias to threat in gSAD.

When attending to threat faces, gSAD relative to healthy controls (HC), exhibit exaggerated activation in rostral ACC (rACC) (Amir et al., 2005, Blair et al., 2008) and subgenual ACC (Goldin, Manber, Hakimi, Canli, & Gross, 2009) indicative of aberrant emotion regulation (Etkin et al., 2011); exaggerated dorsal ACC (dACC) (Phan, Fitzgerald, Nathan, & Tancer, 2006) signifying heightened appraisal or reactivity to threat (Etkin et al., 2011); as well as hyper-activation in key limbic emotion regions (e.g., amygdala, anterior insula; Freitas-Ferrari et al., 2010).

Regarding attentional control, we found less rACC engagement in gSAD relative to HC when attention was directed to shapes in a simple task comprising images of face distractors alongside shapes, but no limbic-related group effects (Klumpp, Post, Angstadt, Fitzgerald, & Phan, 2013). Results suggest a failure to optimally resolve emotional interference in gSAD. Similarly, Blair et al. (2012) observed hypo-activation in the dACC in gSAD participants during an attentional control task also without accompanying differential limbic activation. The dACC is involved in conflict monitoring and action initiation to cognitive demands (Botvinick et al., 1999, Bush et al., 2000, Srinivasan et al., 2013). Thus, a lack of dACC engagement in gSAD indicates a deficiency in controlled cognitive processes.

The general finding of impoverished ACC recruitment in gSAD when top-down control is required is consistent with Attentional Control Theory (ACT), in which the failure of anxious individuals to inhibit task-irrelevant stimuli is due to cognitive efficiency deficits (Eysenck, Derakshan, Santos, & Calvo, 2007). Yet, ACT also proposes anxiety-related impairment can be counteracted when a task is particularly challenging to execute, though at the cost of recruiting more resources. For example, in an emotional Stroop task, high-trait anxious individuals have shown greater dACC activity than low-trait anxious individuals during high-conflict incongruent trials, relative to congruent trials (Krug & Carter, 2010). However, Stroop-related dACC recruitment reflects a late-stage selection process (Silton et al., 2010) and attentional bias to threat in anxious individuals is thought to be somewhat involuntary (Mathews and MacLeod, 1994, Williams et al., 1997) as evinced by the fact that paradigms commonly employed are low in cognitive load (Freitas-Ferrari et al., 2010).

Early stages of attention are modulated by load on attentional resources (O’Connor, Fukui, Pinsk, & Kastner, 2002); thus, varying perceptual load in gSAD may capture ACC responses according to ACT. In a study by Bishop, Jenkins, & Lawrence (2007), load was manipulated to place varying (high/low) demands on attention resources. Anxiety level and dACC activity to fearful face distractors were inversely related in high-trait anxious individuals but only under low load (with a non-significant trend towards the same finding in rACC). Furthermore, state anxiety positively correlated with amygdala response under low, but not high load. In spider-phobia, phobic individuals showed greater amygdala reactivity to distracting spider images than HC regardless of load but no ACC group effects were found (Straube, Lipka, Sauer, Mothes-Lasch, & Miltner, 2011). These mixed results may be due in part to a subclinical sample (Bishop et al., 2007) and perceptual differences when supplanting fearful faces with spider images (Straube et al., 2011).

To our knowledge the modulation of varied perceptual load on ACC in gSAD is not known despite its potential to expand our understanding of attentional bias mechanisms that may not be detected with behavioral measures (e.g., accuracy, reaction time) particularly when compensatory functions occur. Therefore, we employed a paradigm similar to Bishop et al. (2007). Under low perceptual load, we hypothesized that relative to HC, gSAD would show less rACC recruitment and under high load, greater dACC activation. We also explored whether amygdala and/or anterior insula activation to threat distractors would be greater in gSAD to HC.