The Existence of Attentional Biases in Anxiety Disorders

Introduction

As we interact with the world around us, we are presented with immeasurable amounts of visual information from one moment to the next (Boal, Christensen, & Goodhew, 2018). Due to the confines of our perceptual systems, it is impossible to consciously process all incoming information concurrently and only a limited portion of stimuli is attended to at any one time (Kastner & Pinsk, 2004). Attentional mechanisms are therefore crucial in enabling individuals to prioritise and select important information and discard irrelevant or unwanted material (Barry, Vervliet, & Hermans, 2015; Kastner & Pinsk, 2004). One such mechanism is attentional bias to threat, which involves the selective allocation of attention towards emotionally threatening stimuli in the environment relative to neutral or positive stimuli (MacLeod, Mathews, & Tata, 1986; Mogg & Bradley, 1998; Van Bockstaele et al., 2014).

This biased attention to threat-relevant information is a natural result of human evolution (Lucion et al., 2017) and functions essentially as a survival mechanism (Lee & Knight, 2009). The cognitive process allows individuals to quickly detect threatening stimuli in the environment and to prepare adaptive responses in time (Pourtois, Schettino, & Vuilleumier, 2013), which is vital in situations of real danger (LeDoux, 1996). Some research has found that biased attention towards threat can emerge in the early stages of development, with infants as young as 7- and 8-months-old displaying biased attention towards threatening information (LoBue & DeLoache, 2010; Peltola, Leppänen, Palokangas, & Hietanen, 2008). Although primarily innate in nature, it has been suggested that attentional bias towards threat can also develop when incoming stimuli appear to threaten one’s current goals (e.g. Öhman, Flykt, and Esteves, 2001).

Following extensive empirical investigation, it is now widely accepted that anxious individuals exhibit biased attention towards threatening sources of information compared to neutral information, a pattern which is detected to a lesser degree among non-anxious individuals (Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, & van IJzendoorn, 2007; Mogg & Bradley, 1998). Cognitive behavioural models of anxiety advocate that attentional bias is not simply a side-effect or a consequence of emotional distress (Mohlman, Price, & Vietri, 2013). Instead, they propose that the phenomenon plays a causal role in the aetiology and maintenance of anxiety disorders (Beck, Emery, & Greenberg, 1985; Mathews, 1990; Mathews & MacLeod, 2002; Williams, Watts, MacLeod & Mathews, 1988). It is therefore imperative to critically review the research findings on attentional biases in anxiety disorders and corresponding theoretical models as both offer valuable recommendations for how psychological treatments, such as Cognitive Behaviour Therapy (CBT), can be successfully delivered and enhanced for individuals with anxiety.

The current essay begins with a summary of the research evidence demonstrating the existence of attentional biases in anxiety disorders. Next, it describes two specific areas which research has focused on to gain a better understanding of the relationship between attentional biases and anxiety. Thirdly, an overview of some of the methods of assessing attentional bias and relevant findings are provided. Fourthly, the essay outlines associated theoretical models of attentional biases in anxiety which aim to explain how and why attentional biases operate in anxiety. Next, the essay considers the implications that the discussed research evidence and theoretical models have for the delivery and development of CBT for anxiety disorders. Lastly, the essay concludes that there is satisfactory evidence to suggest that research findings on attentional biases do have implications for CBT for anxiety disorders, however further research is warranted.

The existence of attentional biases in anxiety disorders

Over the last three decades, a wealth of research has confirmed the presence of attentional bias towards threat in individuals with various anxiety disorders, including; Generalised Anxiety Disorder (e.g. Beckler, Rinck, Margaraf, & Roth, 2001; Goodwin, Yiend, & Hirsch, 2017; Mogg & Bradley, 2005), Panic Disorder (e.g. Kroeze & van de Hout, 2000; Lim and Kim, 2005; Teachman, Smith-Janik, & Saporito, 2007), Obsessive Compulsive Disorder (e.g. Bradley et al., 2016; Muller & Roberts, 2005), Health Anxiety (e.g. Karademas, Christopoulou, Dimostheni, & Pavlu, 2008; Witthöft, Rist, & Bailer, 2008), Social Anxiety (e.g. Amir, Elias, Klumpp, & Przeworski, 2003; Grafton & MacLeod, 2016), and Post Traumatic Stress Disorder (e.g. Bardeen, Fergus, & Orcutt, 2013; Pineles, Shipherd, Mostoufi, Abramovitz, & Yovel, 2009; Foa, Feske, Murdock, Kozak, & McCarthy, 1991). Indeed, Bar-Haim and colleagues (2007) concluded, based on their meta-analysis of 172 studies, that attentional bias towards threat is prevalent across all anxiety disorders, is observed in children, adults, and older adults, and is displayed by both clinically and non-clinically anxious individuals. With such robust evidence demonstrating the existence of attentional biases in anxiety disorders, researchers hypothesised that such biases may contribute to the development and maintenance of anxiety disorders (Mobini & Grant, 2007).

Understanding the relationship between attentional bias and anxiety

The hypothesis outlined above shifted the focus of research more towards understanding how and why attentional biases occur in anxiety and investigating the mechanisms and components underlying the phenomenon which might contribute to the development and maintenance of anxiety.

One significant question which researchers have endeavoured to answer is whether attentional biases reflect initial, automatic information processing or later, strategic information processing, or indeed if they occur across both stages of processing. Automatic processing refers to processing that does not require cognitive capacity or awareness and is involuntary, effortless, and unintentional in nature (McNally, 1995; Shiffrin & Schneider, 1977). This automatic allocation of attention, often referred to as bottom-up attention, is believed to be moderated by sub-cortical pathways in the brain and is designed to swiftly recognise salient stimuli in the environment (Boal et al., 2018; Davis & Whalen, 2001). In contrast, strategic processing is effortful and intentional, it requires cognitive awareness and conscious attention, and is dependent on voluntary control (McNally, 1995; Shiffrin & Schneider, 1977). This strategic allocation of attention, often referred to as top-down attention (Goodwin, et al., 2017), is understood to be regulated by the prefrontal cortex region of the brain, which governs goal-driven behaviour, attentional control, and working memory (Miller & Cohen, 2001).

A second area of investigation is the underlying components of attentional bias in anxiety (Cisler, Bacon, & Williams, 2009; Fox, Russo, Bowles, & Dutton, 2001). More specifically, research has sought to determine whether attentional biases in anxiety disorders reflect facilitated attention towards threat, delayed disengagement from threat, attentional avoidance of threat, or a combination of these components. (Barry et al., 2015; Cisler & Koster, 2010). Facilitated attention towards threat refers to the rapid and effortless allocation of attention towards pertinent and potentially threatening stimuli in the environment (Cisler & Koster, 2010; Eysenck, Derakshan, Santos, & Calvo, 2007). Delayed disengagement from threat, which typically follows facilitated attention to threat, describes the extent to which a threatening stimulus retains attention and impedes one’s ability to direct attention away from the threat to another stimulus (Cisler & Koster, 2010). Attentional avoidance of threat describes the preferential allocation of attention away from the spatial location of the threatening stimulus to an alternative spatial location (Mogg & Bradley, 2016).

The following section provides an overview of some of experimental methods employed to assess attentional biases in anxiety.

Methods of measuring attentional bias

Several experimental paradigms have been devised to measure attentional bias to threat in anxiety but also to examine its underlying components and mechanisms. These include the modified Stroop task (Stroop, 1935), dot probe task (MacLeod et al., 1986), visual search task (Öhman, et al., 2001), and spatial cueing task (Fox, et al., 2001; Posner, 1980). This essay primarily focuses on the findings from studies using the modified Stroop and dot probe tasks.

Modified Stroop task

In the modified Stroop task (Stroop, 1935), threatening and nonthreatening words are presented in various colours on a computer screen. Participants are asked to identify the colour of the word as quickly as possible, while ignoring the semantic content or meaning of the word. Attentional bias to threat is assumed when participants’ responses are slower or less accurate when naming the colour of a threatening word compared to a nonthreatening word. The general understanding of this Stroop effect is that the threatening content of the word captures attention resulting in less attentional capacity for identifying and naming the colour of the word (van Bockstaele et al., 2014).

Stroop studies have been relatively consistent in demonstrating attentional biases towards threat in clinically and non-clinically anxious populations (for reviews, see Bar-Haim et al., 2007; Mogg & Bradley, 1998; Williams, Mathews, & MacLeod, 1996). However, several critics have challenged the general interpretation that the Stroop effect reflects enhanced attention towards threat. De Ruiter and Brosschot (1994), for example, argued that slower response times could result from cognitive avoidance of threat rather than attentional bias towards threat. Moreover, Algom, Chajut, and Lev (2004) proposed that the processing of fear-related words is likely to increase state anxiety (anxiety experienced in the moment) which would expectedly stall reaction times. According to Cisler and colleagues (2009), the Stroop task does accurately assess facilitated attention or difficulty disengaging but it can be used to measure automatic or strategic processing during attention.

Adaptions of the Stroop task include the masked Stroop task and the unmasked Stroop task, which measure automatic biases and strategic biases respectively. In the masked Stroop task, conscious recognition of the word is blocked by brief presentation (e.g. 17 ms) and rapid replacement of the target stimulus with a backwards mask (e.g. ‘xxxx’) in the same colour. In the unmasked Stroop task, the stimulus is displayed until the participant gives an answer. Several studies have revealed attentional bias for masked threat words among anxious individuals (e.g. Mogg, Bradley, Williams, & Mathews, 1993; Van Den Hout, Tenney, Huygens, Merckelbach & Kindt, 1995) suggesting automatic information processing. However, others have argued that attentional bias reflect strategic information processing based on unmasked Stroop findings (e.g. Thorpe & Salkovskis, 1997).

Dot probe task

One advantage of the dot probe task over the Stroop task is that it measures biased spatial allocation of attention when individuals are asked to make instant responses to threatening stimuli (Iijima, Takano, & Tanno, 2017). In the dot probe task (MacLeod at al., 1986), two stimuli appear briefly at different spatial locations on a computer screen. One stimulus is a threatening word (e.g. “kill”) or picture while the other is an emotionally neutral word (e.g. “chair”) or picture. Both stimuli disappear after 500ms and a dot probe immediately appears in one of the two locations previously occupied by the stimuli. The participant is asked to indicate the location of the probe as soon as it is perceived by pressing a corresponding button. It is expected that responding will be faster when attention is already focused on the spatial location where the probe appears. Attentional bias to threat is assumed when participants respond more quickly to probes presented in the location of this threatening stimulus. This response pattern may result from facilitated attention to threat or delayed disengaged from threat (Bar-Haim et al., 2007), whereas slower responses to probes presented in the location of the emotionally neutral stimulus may indicate attentional avoidance (Mogg & Bradley, 2016).

Several dot probe studies have revealed facilitated attention towards threat across several anxiety disorders, including; GAD (e.g. Bradley, Mogg, White, Groom, & de Bono, 1999) and social phobia (Pishyar, Harris, & Menzies, 2004). However, critics (e.g. Koster, Crombez, Vershuere, and De Houwer, 2004) have suggested that the dot probe task is incapable of measuring facilitated attention towards threat given its methodical flaws, including its failure to measure baseline attention. Some studies using the dot probe have indicated delayed disengagement from threat (e.g. Koster et al., 2004; Salemink, van den Hout, & Kindt, 2007), while others have suggested attentional avoidance (e.g. Chen, Ehlers, Clark, and Mansell, 2002). In terms of information processing, dot probe studies have provided evidence that attentional biases involve both automatic and strategic processing (e.g. Mogg, Bradley, Williams, 1995).

Irrespective of the discrepancies, the above experimental paradigms, as well as the visual search and dot probe tasks, have produced findings which suggest that attentional biases in anxiety operate at both automatic and strategic stages of information processing and they are comprised of several observable components, including; facilitated attention towards threat, delayed disengagement from threat, and attentional avoidance of threat.

Theoretical models of attentional biases in anxiety disorders

Following rigorous empirical investigation, several theoretical models were put forward in recent decades, all of which attempt to explain the relationship between attentional biases and anxiety. One key challenge is the extent of theoretical disagreement in terms of the components and mechanisms underlying the phenomenon. There is a need to better understand the opposing theoretical stances on the relationship between attentional biases and anxiety because each offer differing implications for the psychological treatment of anxiety disorders.

Automatic v’s Strategic Processing

Several theoretical accounts propose that attentional bias in anxiety is solely an automatic process, operating outside of conscious awareness (e.g. Eysenck et al., 2007; Mogg & Bradley, 1998; Williams et al., 1988). These models suggest that incoming stimuli are automatically processed by some form of threat detection mechanism which directs attention towards threatening stimuli. Williams and colleagues’ (1988) model, for example, posits that the threat value of an incoming stimulus is determined by an affective decision mechanism, while Mogg and Bradley’s (1998) model suggests that a valence evaluation system preconsciously appraises incoming stimuli. Only one theoretical account; Wells and Mathew’s (1994) self-regulatory executive function model, postulates that attentional bias in anxiety is exclusively strategic in nature, involving controlled processing and conscious appraisal. According to this model, anxiety is characterised by a consciously perceived threat to self-perseverance and one’s current goals which leads to deliberate monitoring of threat. Importantly, most theoretical models assume that attentional biases operate across both automatic and strategic stages of processing (e.g. Bar-Haim et al., 2007; Beck & Clark, 1997; Öhman, 1996; Matthews and Mackintosh, 1998).

Components of attentional bias in anxiety disorders

Theoretical models advocating automatic attentional biases theorise that facilitated attention towards threat is a key component of the phenomenon (e.g. Bar-Haim et al., 2007; Beck & Clark., 1997; Eysenck et al., 2007; Matthews & Mackintosh., 1998; Mogg & Bradley, 1998; Öhman, 1996; Williams et al., 1988). Beck and Clark’s (1997) information processing model of anxiety, for example, suggests that when an incoming stimulus has been appraised as threatening, a primal threat mode is activated whereby attention automatically orients towards the threat. Difficulty disengaging from threat is a second attentional component considered by several models (e.g. Bar-Haim et al., 2007; Beck & Clark, 1997; Eysenck et al., 2007). Beck and Clark (1997) theorise that when attention is directed towards threat, threat-relevant schema capture and maintain attention. Attentional avoidance of threat is only represented by one model (Williams et al., 1988) which proposes that a resource allocation mechanism channels attention away from threat in low-trait anxious individuals.

In summary, several models of attentional bias in anxiety propose that both initial enhanced attention and subsequent avoidance of threatening stimuli may contribute to the development and maintenance of anxiety disorders.

Implications for Cognitive Behaviour Therapy

CBT is recognised as the gold standard in the psychotherapeutic treatment of anxiety disorders (Otte, 2011) and is recommended by the National Institute for Health and Clinical Excellence for the treatment of various anxiety disorders, including Generalised Anxiety Disorder and Panic Disorder (NICE, 2011), Social Anxiety Disorder (NICE, 2013), and Obsessive Compulsive Disorder (NICE, 2005), in primary, secondary, and community care settings. CBT teaches individuals a range of cognitive and behavioural strategies with the aim of engendering therapeutic change. Cognitive techniques, including Socratic questioning and cognitive restructuring help individuals to alter the content or process of thoughts, cognitive biases, and cognitive schemas, whereas behavioural techniques, such as behavioural activation and graded exposure, strive for changes in observable behaviour (Lorenzo-Luaces, Keefe, & DeRubeis, 2016). The following sections outline the implications for CBT for anxiety disorders based on the research findings and theoretical models discussed above. It is important to note that these implications must be made tentatively due to the obvious inconsistencies across empirical findings and theoretical models.

Implications for CBT based on automatic attentional processing

According to McNally (1995), attentional bias towards threat involves automatic, uncontrollable, and sometimes unconscious processing, and therefore verbally-mediated efforts to modify such biases are likely to be unsuccessful. Behavioural interventions, such as graded exposure and anxiety management training, are likely to be more effective because of their ability to alter fear structures by reducing, and possibly eliminating, maladaptive fear-related associative links and establishing and enhancing new, more adaptive associations (Foa & Kozak, 1986 Tobon, Ouimet, & Dozois, 2011). Mobini and Grant (2007) explain that such behavioural techniques support anxious individuals to develop and exercise mastery over their attentional processing.

Several studies have provided evidence that behavioural therapy alone can have a positive therapeutic effect on attentional biases among individuals with various anxiety disorders. Mathew, Mogg, Kentish, and Eysenck (1995), for example, investigated the effectiveness of a 7-session anxiety management training course in reducing anxiety symptomology and attentional bias among individuals with GAD. The researchers found that both anxiety symptomology and attentional bias were reduced post-intervention. Moreover, Lavy, Van Den Hout, and Arntz (1993) demonstrated that one session of in vivo exposure can reduce attentional bias in individuals with spider phobia. Similarly, Ohman and Soares (1994) suggested that in vivo exposure may directly correct involuntary, unconscious biases in individuals with snake phobia and spider phobia. Furthermore, Reinecke, Waldenmaier, Cooper, and Harmer (2013) observed a rapid decrease in attentional bias among individuals with panic disorder following a single session of exposure therapy, and this post-treatment change in attentional bias predicted further decreases in symptoms four weeks later. Foa and Kozak (1986) explain that exposure therapy encourages anxious individuals to fully engage with the feared situation without resistance or avoidance while allowing themselves to experience the consequent anxiety and fear. The behavioural account is that persistent experiences of a neutral stimulus as conditioned stimulus accompanied with a threatening experience as an unconditioned stimulus results in the allocation of attention towards the conditioned stimulus (Dawson, Schell, Beers, & Kelly, 1982; Mackintosch, 1975). Hence, an automatic attentional bias can be eliminated if the neutral stimulus is presented without the unconditioned reinforcement (Mobini & Grant, 2007).

Implications for CBT based on strategic attentional processing

In line with the view that attentional bias to threat reflects strategic and voluntary processing, cognitive therapy techniques and verbally-mediated interventions may be more effective in targeting attentional biases (Mobini & Grant, 2007). Cognitive models of anxiety disorders emphasise the role of cognitive distortions in the maintenance of the problem. Such cognitive distortions include catastrophic misinterpretations in panic disorder (Clark, 1986) and metacognitive beliefs in GAD (Wells, 1997). These cognitive distortions, combined with the disorder specificity of attentional biases, suggest that thought processing and attentional processing are interlinked (Mobini & Grant, 2007). Cognitive therapy techniques, such as cognitive restructuring, help individuals to identify and challenge threat-related misinterpretations, and to separate feelings of threat from actual threat. This allows individuals to modify negative automatic thoughts and danger schemas, which may in turn reduce or eliminate strategic attentional biases (Mobini & Grant, 2007).

Implications for CBT based on automatic and strategic attentional processing

Although separating automatic and strategic biases has been helpful conceptually, much of the empirical data and theoretical accounts discussed suggest that attentional biases can be both automatic and strategic as opposed to one or the other. Therefore, a combination of behavioural and cognitive therapeutic strategies (i.e. CBT) would help target both voluntary and involuntary attentional biases leading to even greater reductions in anxiety. This is in line with Beck and Clark’s (1997) information processing model, which suggests that the deactivation of the primal threat mode and the strengthening of more reflective and elaborative appraisal of the threatening stimulus is the core task in the treatment of anxiety. Tobon and colleagues’ (2011) systematically reviewed 13 studies which measured the effects of CBT for anxiety on attentional bias and their findings highlight the effectives of CBT in reducing attentional biases. Ten of the 13 studies reviewed demonstrated that attentional bias to threat was significantly reduced following CBT across OCD, GAD, spider phobia, and social phobia. The researchers concluded that both conscious and preconscious attentional biases to threat may be amenable to CBT and that this psychological intervention can reduce, and possibly eliminate, such biases (Tobon et al., 2011).

Implications for CBT based on components attentional bias

Research investigating the observable components of attentional bias (facilitated attention, delayed disengagement, and attentional avoidance) also offer important implications for the treatment of anxiety disorders. It is extremely important that clinicians are aware of these components so that they can tailor treatment in accordance with the client’s presenting problem and their attentional tendencies (i.e. towards or away from threat). Threat-related biases in attention have predominantly been portrayed as negative in the literature and something which needs eradicating to reduce anxiety. In contrast, Barry and colleagues (2015) advise the opposite and suggest that training attention toward or away from threat can in fact facilitate clinical improvement. In social anxiety disorder, for example, self-focused attention prohibits clients from focusing on external information in their surroundings which might invalidate their beliefs about being negatively evaluated during or after social interactions. Like in exposure therapy for panic disorder, it may be important to train socially anxious clients to shift their attention towards the threating stimuli to ultimately reduce their anxiety. In recognising the potential benefits of attentional training, Attention Bias Modification (ABM) treatment is considered below.

Attention-biased modification as an adjunct to CBT

ABM treatment for anxiety, which is based on empirical findings and theoretical models of attentional bias in anxiety, applies computer-based training methods to implicitly manipulate attentional biases to reduce anxiety (Bar-Haim, 2010). ABM typically uses the dot probe task to systematically train participants to avoid threatening stimuli by directing their attention, without their knowledge, to neutral or positive stimuli (Cristea, Kok, & Cuijpers, 2015). The treatment is based on the hypothesis that individuals can only completely attend to one stimuli or thought at any given time (Mobini & Grant, 2007). ABM aims to target automatic attentional processes that have been considered unresponsive to explicit cognitive interventions (Bar-haim, 2010).

Research studies have demonstrated the effectiveness of ABM treatment for a range of anxiety disorders, including GAD (e.g. Amir, Beard, Burns, and Bomyea, 2009; Hazen, Vasey, & Schmidt) and social anxiety (e.g. Schmidt, Richey, Buckner, and Timpano, 2009). Schechner et al., (2014) found that augmenting CBT and ABM treatment resulted in better treatment outcomes compared to CBT alone.  Such results suggest that ABM treatment may be a helpful adjunct to CBT in reducing attentional biases and symptoms of anxiety. However, several meta-analyses examining the effectiveness of ABM treatment found small-to-medium effect sizes for anxiety reduction (e.g. Beard, Sawyer, & Hofmann, 2012; Hakamata et al., 2010; Mogoase, David, & Koster, 2014). These relatively small effect sizes suggest that possible factors which could improve therapeutic effects of ABM need to be considered further (Pergamin-Hight, Naim, Bakermans-Kranenburg, van Ijzendoorn, & Bar-haim, 2015). One such factor which Pergamin-Hight and colleagues (2015) consider important is the nature of stimuli being used in ABM treatment. The researchers suggest that ABM treatment efficacy may be enhanced by using stimuli that are congruent with the content of the anxiety disorder being targeted as it results in stronger activation of underlying neuro-cognitive mechanisms.

Conclusion

Having reviewed the empirical data on attentional bias in anxiety, and associated theoretical models, the research findings do indeed have implications for CBT for anxiety disorders. However, due to the inconsistencies across empirical findings and theoretical models the recommendations for clinical practice must be made cautiously. Nevertheless, automatic attentional biases appear to be more responsive to behavioural interventions, whereas strategic attentional biases seem to be more responsive to cognitive interventions. In terms of the observable components of attentional bias, clinicians should carefully consider the attentional patterns of clients prior to treatment so that the intervention can be tailored in accordance with cognitive profile and presenting problem of the client. Finally, the specific components and mechanisms underlying attentional bias require additional examination to further develop and improve CBT and ABM treatment for anxiety disorders.