Attentional bias and modification of attention bias using graphic health warning labels
In the preceding chapters, the centrality of craving to the psychological maintenance of nicotine addiction (Baker, Breslau, Covey, & Shiffman, 2012) and the relationship that internal and external cues to smoke have with craving elicitation were discussed. An initial study investigating the multiplicative impact of cues suggested that once stress (a commonly reported internal cue) is active, further exposure to smoking-related external cues (such as images of people smoking) does little to further increase craving; once stressed, craving remained elevated even when non-smoking-related cues were presented.
Following this initial study, the remainder of the thesis shifted from a focus on the effect of cues on craving to consider the role of attention and attentional biases in cue-related processing. As was noted (Chapter 2), attentional bias is an important aspect of nicotine addiction that interacts with subjective experience (i.e. craving) to influence smoking behaviour. Attentional bias is associated with clinical outcomes among smokers, with greater bias predicting a greater frequency of use and relapse (Field & Cox, 2008). As importantly, while self-reported craving remains a vital part of craving assessment (Sayette et al., 2000), attentional bias (an objective correlate or index of craving) provides an alternative measure that can be objectively assessed. Importantly, evidence also suggests that attentional bias is sensitive to direct manipulation, offering a promising avenue for the development of treatments targeting reductions in smoking and craving. Despite such promise, however, attentional bias modification investigations have thus far had limited success with most studies reporting reductions in attentional bias but not in smoking outcomes. Although this discrepancy may reflect a weaker link between attentional bias and smoking behaviour, the pattern may also reflect a poor fit between the characteristics of the specific attentional tasks used and the effects the task was intended to have. More concretely, attentional bias modification studies have typically attempted to interfere with cue-related bias by redirecting attention to a neutral cue. Because neutral cues (by definition) do not “grab” attention, such designs are inconsistent with the core function of the attentional system in prioritizing motivationally relevant stimuli and have poor ecological validity. Given this key limitation, an alternative approach that uses the graphic images displayed on nicotine product packaging as a motivationally relevant cue is proposed as likely being better suited to interfering with attentional bias towards smoking cues.
In furthering the basis for this approach to reducing attentional bias, the remainder of this chapter will establish the importance of attentional bias in the elicitation of cue-related craving, outline the clinical consequences of increased attentional bias in smokers, and highlight the limited success of previous assessments of attentional bias modification. The chapter concludes by proposing an approach to attentional bias modification in which avoidance promoting emotions (fear and disgust) are used to compete for attentional resources among smokers.
What is attentional bias
Theories of attention suggest that because the brain cannot process all available information, attentional mechanisms are needed to help “sort” information (or stimuli) (Norman & Shallice, 1986). These mechanisms prioritise the processing of “relevant” stimuli and inhibit the processing of extraneous stimuli, protecting the brain from informational overload (Bargh, 1982). Attentional bias is the term given to instances when the brain appears to prioritize the attentional processing of certain types of stimuli over others (Bar-Haim, Lamy, Pergamin, Bakermans-Kranenburg, & Van Ijzendoorn, 2007). An extensive literature shows that humans (and animals) display an attentional bias toward threat-relevant information, such as snakes, blood, and angry faces, as well as towards appetitive stimuli, such as the availability of food, sex, and reward (Öhman, Flykt, & Esteves, 2001). Preferential processing of goal-related information, such as rewarding and threatening stimuli, and/or stimuli involved in the generation of emotional responses is a core component of the evolved design of the attention systems (MacLeod, Rutherford, Campbell, Ebsworthy, & Holker, 2002).
Making this process more complex is the presence of individual differences in stimulus preference (an index of motivational relevance) or the types of stimuli that are goal-relevant for different persons. For example, hungry versus satiated individuals differentially attend to food cues (Nijs, Franken, & Muris, 2008; Nijs, Muris, Euser, & Franken, 2010; Stockburger, Schmälzle, Flaisch, Bublatzky, & Schupp, 2009), high versus low health anxious persons respond differently to health threat cues (Lees, Mogg, & Bradley, 2005; Owens, Asmundson, Hadjistavropoulos, & 38 Owens, 2004), chronic pain groups preferentially attend to pain sensory and pain affective stimuli (Roelofs, Peters, Zeegers, & Vlaeyen, 2002), and persons with a substance use disorder preferentially attend to substance-related cues (Field & Cox, 2008; Littel, Euser, Munafò, & Franken, 2012).
As was noted in Chapter 2, a preoccupation with drug-related cues is generally seen as important in maintaining craving and addiction; this preference has been conceptualized as an attentional bias. Attentional bias to drug-related cues is thought to have causal effects on substance abuse, addiction development, and the maintenance of addictions (Field & Cox, 2008; Franken, 2003; T. E. Robinson & Berridge, 1993; T. E. Robinson & Berridge, 2000; Weinstein & Cox, 2006). In the context of nicotine addiction, attentional bias refers to the observation that smoking-related cues tend to “grab” smokers’ attention, increase craving, and maintain cigarette consumption (Field & Cox, 2008). Similar patterns of attentional bias are evident both in studies of nicotine dependence as well as in studies of alcohol (Jones, Bruce, Livingstone, & Reed, 2006; Stetter, Ackermann, Bizer, Straube, & Mann, 1995), cannabis (Field, Mogg, & Bradley, 2005), heroin (Lubman, Peters, Mogg, Bradley, & Deakin, 2000) and cocaine (Vadhan et al., 2007) dependence. Such data attest to the widespread effects attentional bias has across types of addiction.
Attentional bias in smokers
Several comprehensive reviews have summarized the literature surrounding attentional bias for substance-related stimuli (presented verbally, pictorially, or as direct exposure) in both nicotine users and other substance use populations (W. M. Cox, Fadardi, & Pothos, 2006; Field, 2006; Field & Cox, 2008; Field, Munafò, & Franken, 2009; Franken, 2003; Littel et al., 2012; Robbins & Ehrman, 2004). When assessing attentional bias in smokers, researchers often make use of our finite processing capacity by designing/using tasks that contain two attentional elements – one that is required to complete the task (representing voluntary attentional processing) and one that intrinsically grabs attention (involuntarily or automatically). A variety of paradigms have been developed, including both direct and indirect measures of attentional bias. Indirect measures include those that are thought to index attentional bias based on participants’ impaired performance on a primary task in the presence of smoking-related cues (such as words or pictures) (Field & Cox, 2008). More direct measures of attentional focus on smoking-related stimuli include eye-tracking (Baschnagel, 2013; Field, Mogg, & Bradley, 2004; Kang et al., 2012; Kwak, Na, Kim, Kim, & Lee, 2007; Mogg, Field, & Bradley, 2005) and cognitive processing measures of attention, such as event-related potentials (ERPs) using electroencephalography (EEG) techniques (Littel et al., 2012). A summary of this literature, as well as an argument for why ERPs are seen as among the most accurate measures of attentional bias is provided below.
The addiction Stroop is the most widely used test of smoking-related attentional bias. When completing this task, smokers, but not non-smokers, are slower at colour naming smoking-related versus neutral words (Munafò, Mogg, Roberts, Bradley, & Murphy, 2003). However, while it is widely used, the indirect and inferential nature of this measure allows for competing interpretations regarding why colour-naming is slower for smoking-related words. First, rather than indicating that smokers are attracted to smoking-related stimuli the task interference may represent smokers trying to avoid processing smoking stimuli, particularly smokers who are attempting to abstain and avoid cues to smoke (Field & Cox, 2008; Klein, 2007). Alternately, a generic slowdown in cognitive processing as a consequence of craving being induced, rather than selective attention, may lead to slower reaction times (Algom, Chajut, & Lev, 2004). Taken together, although the smoking Stroop task remains the most common paradigm used to assess attentional bias, there are interpretive limitations that accompany this indirect measure of attentional bias that should not be ignored.
Visual probe tasks (VPTs) have also been used to more directly assess how visuo-spatial attention is allocated toward smoking stimuli. As with other methods, studies utilizing VPT methods also suggest greater attentional bias (faster reaction time) toward smoking-related pictorial stimuli in smokers compared to non-smokers (Bradley, Field, Mogg, & De Houwer, 2004; Ehrman et al., 2002; Field, 2006; Mogg, Bradley, Field, & De Houwer, 2003). However, reaction time to VPTs only provides information about the stimuli that is being attended to at the time of offset and attentional bias results vary when pictures are presented for varied amounts of time (Field, Mogg, Zetteler, & Bradley, 2004; Noël et al., 2006). In other words, reaction times may index any number of attentional processing stages, such as initial orienting, maintenance, or disengagement of attention (Allport, 1989; Field & Cox, 2008; LaBerge, 1995). These diverse elements of attentional processing are not equivalent nor are the interpretations regarding possible smoking processes and cessation they generate.
Improving on the limitations of reaction time data, VPTs have also been used to assess attentional bias using eye-tracking techniques (Baschnagel, 2013; Field et al., 2004; Kang et al., 2012; Kwak et al., 2007; Mogg et al., 2005). In these studies, participants’ eye movements are monitored while smoking-related and neutral cues compete for attention. In general, results from such studies indicate that smokers, but not non-smokers, maintain their gaze on smoking-related stimuli longer than neutral stimuli. Similarly, greater initial fixation to smoking-related stimuli is greater among more dependent smokers (Bradley, Mogg, Wright, & Field, 2003; Mogg et al., 2003) and smokers deprived of nicotine maintain their gaze for longer (Field et al., 2004). While the assessment of continuous eye movements may be closer to directly indexing attention than reaction times, interpretations are still contingent on the assumption that smokers are paying attention to the stimuli their eyes are oriented towards.
Finally, a few studies suggest greater attentional bias toward smoking-related stimuli among smokers when indexing attentional bias using electro-physiology. Relative to reaction time and eye-tracking, electrophysiological measures, such as event-related potentials (ERPs), are among the most direct means of assessing attentional bias because they directly assess neural activation as it occurs. ERPs offer excellent temporal resolution and have the capability to index small changes in attention that might not be detected by reaction time or eye-tracking. Because they are manifestations of neural activity that occur in preparation for, or in response to, discrete events (Fabiani, Gratton, & Coles, 2000), ERPs can provide a more comprehensive understanding of motivated attentional processing.
Two ERPs in particular, the P300 (P3) and the late positive potential (LPP), have been associated with attentional processes and are larger in response to smoking-related stimuli (compared to neutral cues) and in smokers compared to non-smokers (Littel et al., 2012). The P3 is generally thought to reflect the deployment of attentional resources to task-relevant stimuli (Donchin, 1981; Johnson, 1986; Polich & Kok, 1995; Polich, 2007) and is most commonly elicited through paradigms in which participants have to respond to non-frequent (oddball) stimuli presented among a series of frequent, non-target stimuli. In such studies the P3 tends to be enhanced in responses to target relative to non-target stimuli, suggesting increased neural processing (Donchin, 1981; Johnson, 1986). Conversely, the P3 is often decreased when attention is shifted elsewhere, for example when a third element of the paradigm provides a distraction (Duncan-Johnson & Donchin, 1977; Hillyard, Hink, Schwent, & Picton, 1973).
More importantly, in terms of the current thesis, research has also repeatedly shown that the P3 is enhanced in response to motivationally relevant stimuli (those signalling reward or threat) compared to neutral stimuli (Ferrari, Codispoti, Cardinale, & Bradley, 2008; Hajcak, MacNamara, & Olvet, 2010; Schupp et al., 2007). Multiple studies have found that smokers produce larger P3 responses to smoking-related images than to neutral images (Littel & Franken, 2007; McDonough & Warren, 2001; Warren & McDonough, 1999). Supplementing such studies are other recent works examining the LPP, which reflects a continuation of attentional processing of emotional and/or motivationally relevant stimuli (Foti & Hajcak, 2008; Foti, Hajcak, & Dien, 2009; Hajcak, Dunning, & Foti, 2007). The LPP represents a less transient (in comparison to the P3) ERP component (lasting for up to several seconds after stimuli onset) that can also be produced in an oddball paradigm and is thought to be sensitive to top-down cognitive modulation (Hajcak et al., 2010). Studies among smokers have found an enhanced LPP response to smoking-related cues compared to neutral cues (Versace, Minnix et al., 2011) as well as greater LPP responses in former versus never smokers (Littel & Franken, 2007; J. D. Robinson et al., 2015).
Interestingly, studies employing both direct and indirect measures of attentional bias have shown that the size of ERP responses may be moderated by the level of nicotine dependence. Indeed, a significant number of studies utilizing both smoking-Stroop and VPT methods (Bradley et al., 2003; Hogarth, Mogg, Bradley, Duka, & Dickinson, 2003; Mogg et al., 2005; Mogg & Bradley, 2002; Munafò et al., 2003; Waters et al., 2003; Waters & Feyerabend, 2000; Zack, Belsito, Scher, Eissenberg, & Corrigall, 2001) as well as ERP methods (Littel & Franken, 2011; Versace et al., 2011; Versace, Engelmann et al., 2011) have found that the degree of attentional bias towards smoking-related stimuli is moderated by dependence.
Complicating the interpretation of these data, is the fact that the direction of the relationship between attentional bias and dependence (what the “moderation” looks like) appears to differ across indirect and direct measures of attentional bias. In general, indirect measures (although there are contradicting reports; (Mogg & Bradley, 2002; Zack et al., 2001) suggest that more depend smokers display reduced attentional bias toward smoking cues compared to those with lower dependence (Bradley et al., 2003; Hogarth et al., 2003; Mogg et al., 2005; Waters et al., 2003). Conversely, studies utilizing more direct measures of attentional bias (ERP studies) tend to suggest greater attentional bias among more dependent smokers (Littel & Franken, 2011; Versace et al., 2011; Versace et al., 2011). Complexity notwithstanding, it is clear that the degree of attentional bias varies across smokers with differing levels of dependency and that more research is needed to fully understand the relationship between attentional bias and dependence.
In summary, there is clear evidence that smokers display attentional bias toward smoking cues, although these effects tend to vary somewhat as a function of measurement directness and interpretive power. While no measure is without its limitations, ERPs allow perhaps the most direct measure of attentional bias while also providing some confidence in the motivational significance of smoking cues to smokers; although more research is needed to better understand the significance of attentional biases in those with varying levels of dependence.
Clinical consequences of attentional bias
The theoretical models described in Chapter 2 suggest that attentional bias plays a causal role in triggering nicotine use (Field & Cox, 2008). Multiple studies indicate that greater attentional bias to nicotine cues is associated with greater subjective craving (Field et al., 2009; Mogg et al., 2005; Mogg et al., 2003; Mogg & Bradley, 2002; Zack et al., 2001) and a higher frequency of tobacco use (Mogg & Bradley, 2002; Zack et al., 2001). Studies have also shown that experimentally increasing craving through nicotine deprivation (Field et al., 2004; Gross, Jarvik, & Rosenblatt, 1993; Waters & Feyerabend, 2000), a priming dose of alcohol (Burton & Tiffany, 1997; Field et al., 2005), or exposure to environmental smoking cues (holding a lit cigarette and completing a smoking Stroop task) (Field, Rush, Cole, & Goudie, 2007), increases attentional bias. One study (Field et al., 2007) also found that the effects of cue exposure on attentional bias were partially mediated by the effects of cue exposure on subjective craving. Other reports have suggested greater attentional bias is related to a higher frequency of cigarette consumption (Mogg Bradley, 2002; Zack et al., 2001); although there are conflicting reports (Bradley et al., 2003; Hogarth et al., 2003; Mogg et al., 2005; Munafò et al., 2003; Waters et al., 2003; Waters & Feyerabend, 2000). Finally, evidence suggests attentional bias can predict nicotine use or relapse in those trying to abstain from use, such that greater attentional bias on the first day of a quit attempt is associated with shorter time till relapse (Waters et al., 2003). Although these are clearly complex processes, the available evidence tends to suggest that attentional bias is an important correlate of craving and likely contributes to the maintenance of cigarette consumption.
Attenuating attentional bias to reduce smoking
Since establishing that attentional bias is associated with craving and is likely implicated in the maintenance of addiction, research has turned toward investigating whether trying to reduce attentional bias has clinical utility and can be used to improve cessation outcomes. Several studies have assessed whether training smokers to avoid smoking cues (using a VPT) can reduce attentional bias and, in turn, reduce craving and improve behavioural outcomes (Attwood, O’Sullivan, Leonards, Mackintosh, & Munafò, 2008; Begh et al., 2015; Field, Duka, Tyler, & Schoenmakers, 2009; Kerst & Waters, 2014; Lopes, Pires, & Bizarro, 2014; McHugh, Murray, Hearon, Calkins, & Otto, 2010; Wittekind, Feist, Schneider, Moritz, & Fritzsche, 2015). While some studies have failed to find an effect for attentional bias training (Begh et al., 2015; McHugh et al., 2010), most studies suggest that attentional bias itself can be reduced through training (Attwood et al., 2008; Field et al., 2009; Kerst & Waters, 2014; Lopes et al., 2014).
Troublingly, however, reductions in attentional bias do not appear to translate into reduced craving or altered smoking behaviour (Attwood et al., 2008; Field et al., 2009; Lopes et al., 2014). Put simply, reducing attentional bias to smoking cues in smokers does not appear to result in reductions in subjective craving or in smoking behaviour itself. More rigorous training programs have had only modestly better success in reducing craving (Kerst & Waters, 2014), or were conducted among samples already seeking cessation help and/or in which an active control group able to rule out the confounding effects of quit motivations were not included (Wittekind et al., 2015). Taken together, it appears that using a modified version of the VPT can decrease attentional bias toward smoking stimuli but that such changes do not consistently translate to reductions in smoking or craving.
Perhaps one reason attentional bias modification studies have failed to produce changes in clinical outcomes is related to the indirect nature of VPT-based assessment. As mentioned, one weakness in the interpretation of VPTs is that it is not always clear which element of the attentional process the VPT is indexing. It is possible that attention is still initially oriented toward smoking stimuli but, through conditioned learning during the task, the participant learns to quickly reorient their attention to the neutral stimuli. If this is the case, it may be that attentional bias was not actually modified but, rather, participants simply learned to more rapidly divert attention to the neutral cue as part of the task. One way of testing this concern would be to include eye tracking measures to assess the eye movements and gaze duration during the training session. As noted, however, measuring attentional processing via neural metrics, such as EEG would not only allow more confidence regarding the direct assessment of the attentional processing but also be more sensitive to subtle temporal changes in attentional processing across the course of the training.
Another reason previous studies attempting to reduced attentional bias may have failed to produce changes in clinical outcomes is due to the nature of the training itself. In all the previously mentioned studies, participants were trained, via a VPT study design, to preferentially attend to a neutral stimuli and ignore the smoking stimuli (Attwood et al., 2008; Begh et al., 2015; Field et al., 2009; Kerst & Waters, 2014; Lopes et al., 2014; McHugh et al., 2010); in most, attention toward the smoking stimuli was reduced (Attwood et al., 2008; Field et al., 2009; Kerst & Waters, 2014; Lopes et al., 2014). However, it is unlikely that increasing attention to a neutral cue would possess the motivational impetus to translate to situations outside of the training environment and such changes may more accurately reflect a conditioned learning response. Fundamentally, the VPT fails to take into account the motivational nature of the attentional system. Recall, attentional mechanisms are designed to help ignore irrelevant information and devote attentional resources toward motivationally salient stimuli (P. J. Lang & Bradley, 2010; Öhman et al., 2001). Consequently, it seems plausible that exposing smokers to cues that compete for (and thus interfere with) motivated attentional processing of smoking cues is more likely to be successful (i.e. will lead to reduced attentional bias, smoking craving, and cigarette consumption) than will exposure to a neutral cue.
Table of Contents
TABLE OF CONTENTS
LIST OF TABLES
LIST OF FIGURES
LIST OF ABBREVIATIONS
2.2. DEFINITION AND CLINICAL RELEVANCE OF CRAVING
2.3. DEVELOPMENT OF CUE-INDUCED CRAVING
2.4. MULTIDIMENSIONAL ASSESSMENT OF CRAVING
2.5. CLARIFYING THE NATURE OF CRAVING-ELICITING CUES
3. CONTRASTING CRAVING IN RESPONSE TO MULTIPLE SMOKING CUES
3.3. INTRODUCTION .
4. ATTENTIONAL BIAS AND MODIFICATION OF ATTENTION BIAS USING GRAPHIC HEALTH WARNING LABELS
4.2. WHAT IS ATTENTIONAL BIAS
4.3. ATTENTIONAL BIAS IN SMOKERS
4.4. CLINICAL CONSEQUENCES OF ATTENTIONAL BIAS
4.5. ATTENUATING ATTENTIONAL BIAS TO REDUCE SMOKING
4.6. GRAPHIC HEALTH WARNING LABELS
5. USING DISGUST AND HEALTH ANXIETY INDUCING GHWLS TO ATTENUATE MOTIVATED ATTENTIONAL PROCESSING OF SMOKING CUES
5.7. SUPPLEMENTARY ANALYSIS
6. ATTENTIONAL PROCESSING OF GHWLS THAT ELICIT PRIMARILY DISGUST OR HEALTH ANXIETY
7. DISCUSSION .
7.2. SUMMARY OF KEY FINDINGS
7.3. IMPLICATIONS OF THIS RESEARCH WITH RESPECT TO THE CURRENT LITERATURE
7.5. FUTURE DIRECTIONS
7.6. OVERALL CONCLUSIONS
GET THE COMPLETE PROJECT
The impact of stress, health anxiety and disgust on cue-induced craving and attentional bias in smokers