Comparing attentional bias to smoking cues in current smokers, former smokers, and non-smokers using a dot-probe task

Abstract

Much evidence documents that individuals with emotional and drug-use disorders demonstrate biased attention toward stimuli associated with their disorder. This bias appears to diminish following successful treatment. Two studies examined whether current cigarette smokers show biased attention toward smoking-related images compared with non-smokers (Studies 1 and 2) and whether this bias is less pronounced in former smokers (Study 2). Attentional bias toward cigarette-related photographs was examined using the dot-probe task. Pairs of images (one smoking-related) appeared side by side for 500 ms on a computer screen prior to the presentation of a probe (an asterisk) replacing one of the photographs. Subjects struck a key as quickly as possible to indicate the probe location. Attentional bias was defined as faster reaction times when the probe replaced the smoking-related image. In both studies, current smokers displayed significantly greater attentional bias toward cigarette stimuli than did non-smokers. Former smokers in Study 2 displayed an intermediate level of bias, but did not differ significantly in bias score from either of the other groups. These results support further use of the dot-probe task as a measure of attentional bias in non-abstinent smokers and in individuals undergoing smoking cessation treatment.

Introduction

A great deal of research now documents that events related to drug use come to evoke drug-related responses such as high, withdrawal, craving, and arousal (see Robbins and Ehrman, 1992, Drummond et al., 1995, O'Brien et al., 1998 for recent reviews). Many theorists view drug-related cues as a causal factor in drug use and relapse to drug use following treatment (Wikler, 1965, Siegel, 1979, Stewart et al., 1984, Childress et al., 1986, Baker et al., 1987, Tiffany, 1990). In the specific case of smoking, cigarette-related cues have been shown to both increase craving and decrease the latency to smoke (e.g. Abrams et al., 1988, Niaura et al., 1988, Niaura et al., 1992, Droungas et al., 1995). Although most discussions of cue reactivity have not explicitly addressed the role of attention to drug cues, the incentive-sensitization model of Robinson and Berridge (1993) postulates that drug-related stimuli become more salient or attention-grabbing as a consequence of chronic drug use. Such a process could produce a positive feedback loop that magnifies the risk posed by drug-related cues. That is, enhanced attention to drug-related cues should produce enhanced responding to those cues. Increases in cue-induced responses may promote greater drug use that in turn increases drug cue salience.

Although relatively little effort has been devoted thus far to investigating attentional processes in substance abuse, many studies have examined the role of attention in other emotional disorders. Much of this research has made use of the emotional Stroop color-naming task. In the Stroop task, individuals are asked to call out as quickly as possible the color of the ink in which target words appear. Greater interference with color naming (increased verbal response times) is taken as evidence for enhanced attention to the meaning of the target words. In two extensive literature reviews, Williams et al., 1996, Williams et al., 1997 discuss studies demonstrating that individuals with a variety of emotional disorders including depression, generalized anxiety disorder, panic disorder, phobias, and PTSD show increased color-naming times when the target words are disorder-relevant. Two findings in this literature are particularly striking. First, color-naming interference is highly selective; individuals show impairments only for words specifically related to their disorder (e.g. McNally et al., 1990, Watts et al., 1986). Second, successful treatment reduces color-naming interference (Watts et al., 1986, Gotlib and Crane, 1987, Mattia et al., 1993, Mathews et al., 1995). Thus, Stroop interference by affect-laden words appears to be a specific index of an individual's current level of preoccupation or distress.

Several studies in recent years have extended the emotional Stroop task to studies of smokers. Gross et al. (1993) found that active smokers who are asked to abstain from smoking for 12 h before a test session show greater Stroop interference with cigarette-related words than with control words. Similar results were reported by Waters and Feyerabend (2000) using smokers who were deprived of cigarettes for 24 h prior to testing. However, these studies did not find interference differences between smoking-related words and control words in non-abstinent smokers and did not test non-smokers. A third study (Johnson et al. 1997) compared current smokers, recent quitters, and non-smokers and failed to find an overall effect of group on Stroop performance. Most recently, Wertz and Sayette (2001) examined Stroop performance in smokers who were told that they would or would not be allowed to smoke during the experiment. Individuals given an expectation of smoking showed a greater degree of color-naming interference with smoking words. However, all subjects underwent testing following 12 h of smoking deprivation. Thus, attentional bias effects in non-abstinent smokers remain to be demonstrated in the Stroop task.

A second test used to study attentional bias is the dot-probe task first introduced by MacLeod et al. (1986). Subjects are shown two images (pictures or words) side-by-side on a computer screen. When the two images disappear, a target stimulus (the dot probe) appears in place of one of the images. Subjects are asked to indicate the location of the target by striking a key as quickly and accurately as possible. Faster reaction times (RTs) when the target replaces a particular class of events indicate biased attention toward those events. The dot-probe task has been used successfully to demonstrate that individuals with mood disorders such as anxiety or depression shift attention toward events related to those states (MacLeod et al., 1986, MacLeod and Mathews, 1988, Mogg et al., 1992, Mogg et al., 1994, Mogg et al., 1995, Mogg et al., 1997). Induced mood states can have similar effects; hungry individuals (who do not suffer from an eating disorder) show bias toward food-related stimuli (Mogg et al. 1998). Finally, and of most interest here, selective attention effects have also been shown in users of opiates (Lubman et al., 2000) and alcohol (Townshend and Duka, 2001) when presented with pictures of drug-related stimuli in the dot-probe task. The present study was designed to extend the use of the dot-probe task to the study of smokers.

Because cue-induced craving is thought to play an important role in smoking, biased visual attention toward cigarette-related cues may increase the motivation to smoke. Smokers may be more likely than non-smokers to notice in passing an ashtray or pack of cigarettes, thereby increasing their likelihood of experiencing cue-induced craving. The dot-probe task appears better designed than the Stroop task to study such shifts in the direction of visual attention for several reasons.

First, while the Stroop task requires subjects to attend to a single target word, the dot-probe task explicitly examines shifts in attention between two co-present visual events. In this way, the dot-probe task better simulates real-world conditions in which both smoking-related and non-smoking-related objects compete for visual attention. Second, color-naming interference on the Stroop task could result from processes other than initial shifts in attention. For example, words associated with an individual's area of concern might produce greater interference than control words because they produce cognitive/mood states which compete with the color-naming response (e.g. Williams et al., 1996). By contrast, the dot-probe task measures biased attention as the facilitation of target location performance when the target and smoking picture appear in the same position. Targets that appear in place of smoking pictures cannot be responded to faster unless attention has already been directed toward those pictures. Finally, Stroop studies have only produced interference effects in smokers who abstained from smoking prior to the test session. We hoped that the dot-probe task would prove to be a more sensitive measure of biased attention in non-abstinent smokers.

The two studies reported here compare attentional bias to cigarette-related photographs across groups with different smoking histories. Study 1 represented a small pilot study examining whether current smokers show greater attentional bias to smoking-related photographs than non-smokers. Study 2 represented a replication of Study 1 with larger and more diverse samples and the inclusion of a group with a prior smoking history (former smokers). As noted earlier, several Stroop task studies have demonstrated that individuals with emotional disorders show reduced Stroop interference by disorder-related words following successful treatment (Watts et al., 1986, Gotlib and Crane, 1987, Mattia et al., 1993, Mathews et al., 1995). By analogy, we predicted that former smokers would show less bias toward smoking-related events than current smokers.