Cognitive Bias and Social Anxiety Disorder

Today a new study was accepted for publication in Internet Interventions. The paper was about social anxiety disorder (SAD) and attentional bias.

Typically, the visual-attention system is selectively biased towards stimuli of biological importance, such as cues of threat (predators, dangerous individuals) and reward (food, mates) (Frewen et al. 2008). These attentional processes are considered to have a fundamental role to play in the maintenance of social anxiety (SAD) and other anxiety disorders (Bar-Haim et al. 2007). The understanding that attentional processes may be modified using attention bias modification (ABM) training (MacLeod et al. 2002) with a commensurate reduction in clinical presentation of anxiety (Amir et al. 2009; Schmidt et al. 2009) has generated strong interest in the area (Kuckertz & Amir 2015).

Cognitive models of SAD tend to specify a two-stage theory of attentional bias, an initial automatic threat-detection system that orients an individual in response to danger, followed by a conscious, voluntary system that can maintain or override attention (Cisler & Koster 2010). In accordance with cognitive models, those with SAD will tend to bias attention not only towards biological risks but also threatening social information, such as negative facial expressions of nearby individuals, or internal emotional and physical disequilibrium, such as the shaking of their own hands (Boettcher, Leek, et al. 2013a). Persistent negative and distorted views of social situations may reinforce an attention bias towards threats (Beck & Clark 1997).

Attentional bias can be measured using the dot-probe task, by timing the responses of subjects to threatening, neutral and positive images (normally faces) or words displayed on a screen. A typical example of this task has subjects shown two words or images for a short period of time (e.g., 500 ms), after which one of two possible probes appear behind one of the images. Typically the probe is the letter E or F, or one or two dots. The subject must identify the location and differentiate the probe type and then respond by pressing the corresponding button on a mouse or keyboard. Probe placements are balanced between neutral, negative and positive images or words and mean-reaction times for stimuli of each emotional valence are compared to the other. Attentional bias towards threat (hypervigilance) is determined when response times are shorter to probes placed behind threatening stimuli as compared to a neutral or positive stimuli. This would indicate that the subject was drawn to threatening stimuli over neutral stimuli. The opposite result (attentional avoidance) would indicate a subjects turning away from negative stimuli.

[lightbox link=”” thumb=”×924.jpg” width=”1024″ align=”left” title=”Presentation of the attention bias measuring process.” frame=”true” icon=”image” caption=”Presentation of the attention bias measuring process.”]

Alternative tasks for measuring attentional bias have included the modified Stroop, which compares response times to color identification of threatening and neutral words (Andersson et al. 2006); the spatial cueing task in which rectangles on either side of a fixation point are illuminated with a neutral or threatening image, after which a target is presented and subject response time measured; or the visual search task, for which participants are asked to identify a threatening or neutral word within a matrix of rows and columns of the opposite valence (Cisler & Koster 2010). For example, the word cancer may be embedded in a 5×5 matrix of distracting neutral words such as table or vice-versa. The dot-probe task was developed to overcome limitations of the modified Stroop task (Bar-Haim et al. 2007) and has become the most common experimental paradigm used, with hundreds of studies to date (Price et al. 2014).

According to a meta-analysis by Bar-Haim et al. (2007) the attention bias effect is quite prevalent in anxious populations at a moderate effect size (d = 0.45). Studies using the dot-probe have identified both hypervigilance and attentional avoidance at various stimuli presentation times. Boettcher, Leek, et al. (2013a) have reviewed the evidence. Hypervigilance at 500 ms is currently understood to be the predominant form of attentional bias in individuals with anxiety disorders (Asmundson & Stein 1994; Musa et al. 2003; Mogg & Bradley 2002; Helfinstein et al. 2008; Klumpp & Amir 2009; Mogg et al. 2004). Fewer studies have identified hypervigilance at very fast presentation times of 1000 ms) have not been successful at identifying hypervigilance (Asmundson & Stein 1994; Musa et al. 2003; Mogg et al. 2004; Helfinstein et al. 2008). A few studies have identified attentional avoidance at 500 ms (Chen et al. 2002; Vassilopoulos 2005). Supplementary verification using eye-tracking studies have verified both hypervigilance (Schofield et al. 2012; Wieser et al. 2009) and attentional avoidance (Wieser et al. 2009; Mühlberger et al. 2008). There has also been success identifying attentional bias using the modified Stroop task (Linnman et al. 2006), visual search (Juth et al. 2005), and spatial cueing tasks (Bar-Haim et al. 2007).

Multiple experimental paradigms showing evidence of attentional bias reduce the chance that it is merely an artifact of a particular paradigm (Cisler & Koster 2010). Nevertheless, some studies have shown no attentional bias effect at similar time periods (Mohlman et al. 2013; Waters et al. 2004). The results are particularly poor for internet-based studies (Boettcher, Leek, et al. 2013a). The sometimes conflicting results for both attentional bias and the reduction of anxiety during ABM training (Hakamata et al. 2010; Hallion & Ruscio 2011; Andersson et al. 2005) have led some to refer to ABM as “the emperor’s new clothes” (Emmelkamp 2012). Others such as Clarke et al. (2014), have drawn the conclusion that further experimentation and “scrutiny into the precise task conditions and modes of delivery” (p. 4) is needed.

Certain presentations of the dot-probe task have shown stronger effects than others. Two meta-analyses have suggested that using top and bottom placement of images has better results than placing images side-to-side (Hakamata et al. 2010; Beard et al. 2012). Results have tended to be more reliable when the dot-probe has a bottom placement rather than a top placement as subjects might preferentially reference the top image despite its valence (Price et al. 2014). Dot-probe studies that presented words rather than faces have shown stronger effects, despite the suggestion that it is less ecologically valid (Hakamata et al. 2010; Beard et al. 2012). The type of facial stimuli (cf., Samuelsson et al. 2012) used may also be a factor. Boettcher (2012) in a program delivered online at home, failed to find attentional bias while carrying out methodologies similar to those that found strong effects (Amir et al. 2009; Schmidt et al. 2009), but using different facial stimuli.

Alteration of attentional bias towards threatening stimuli using ABM training has resulted in success reducing clinically significant anxiety and created strong interest in the area. A recent review by Kuckertz & Amir (2015) identified 231 studies via PsychINFO using the search terms “attention bias modification” OR “attention training” OR “attention modification,” with over half in the last 3 years. However, while ABM treatment appears promising, the reliability of measuring attentional bias in internet administered studies has proven problematic and requires further exploration (Kuckertz & Amir 2015).

The present study

The purpose of this study was to measure attentional bias toward threatening, neutral and positive images or words using an Internet administered dot-probe task and to identify to what degree this bias correlated with a variety of outcome measures. Subjects in the study were all diagnosed with SAD as identified by the Liebowitz social anxiety scale or LSAS-SR (Liebowitz 1987). Outcome measures included the Quality of Life Inventory scale (QOLI; Frisch et al. 1992), the mini Social Phobia Inventory (Mini-SPIN; Connor et al. 2001) the General Anxiety Disorder scale (GAD-7; Spitzer et al. 2006) and the Patient Health Questionnaire scale (PHQ9; Kroenke et al. 2001).

The results

The purpose of this study was to measure pre-treatment attentional bias in 153 participants diagnosed with SAD using a web-based version of the dot-probe paradigm. Results showed no significant correlation for attentional bias (towards or away from negative words or faces) and the self-rated version of the Liebowitz Social Anxiety Scale (LSAS-SR). However, two positive correlations were found for the secondary measures Generalized Anxiety Disorder 7 (GAD-7) and Patient Health Questionnaire 9 (PHQ-9). These indicated that those with elevated levels of anxiety and depression had a higher bias towards negative faces in Neutral-Negative and Positive-Negative valence combinations, respectively. The unreliability of the dot-probe paradigm and home-based internet delivery are discussed to explain the lack of correlations between LSAS-SR and attentional bias. Changes to the dot-probe task are suggested that could improve reliability.

You can read the full paper here (in a few days):

Miloff, A., Savva, A., & Carlbring, P. (in press). Cognitive Bias Measurement and Social Anxiety Disorder: Correlating Self-Report Data and Attentional Bias. Internet Interventions. doi: 10.1016/j.invent.2015.03.006