Psychopharmacology

, Volume 194, Issue 2, pp 279–288

Effect of impulsivity on craving and behavioral reactivity to smoking cues

Authors

    • University of Illinois at Chicago
    • Department of PsychiatryUniversity of California
  • Bonnie Spring
    • University of Illinois at Chicago
    • Hines VA Hospital
    • Department of Preventive Medicine, Feinberg School of MedicineNorthwestern University
  • Dennis McChargue
    • University of Illinois at Chicago
    • Hines VA Hospital
    • Department of PsychologyUniversity of Nebraska—Lincoln
Original Investigation

DOI: 10.1007/s00213-007-0832-x

Cite this article as:
Doran, N., Spring, B. & McChargue, D. Psychopharmacology (2007) 194: 279. doi:10.1007/s00213-007-0832-x

Abstract

Introduction

Nearly 25% of American adults remain regular smokers. Current smokers may be especially likely to possess characteristics that impair their ability to quit, such as impulsivity. Impulsive individuals may be overly prone to smoke because they are particularly drawn to rewarding stimuli and related cues. The aim of this study was to test the hypothesis that more impulsive smokers are more responsive to cigarette cues than other smokers.

Materials and methods

In a repeated measures design, 60 euthymic, adult smokers (50% female) were exposed to a smoking cue and a neutral cue in two experimental sessions. Cue reactivity was operationalized as changes in cigarette craving and preference for immediate vs delayed smoking after cue exposure.

Results

Impulsivity predicted a heightened craving response to both cues but particularly the smoking cue (t [161] = 3.21, p = 0.002). Smokers with high levels of impulsivity exhibited a greater preference for immediate rewards over larger, delayed rewards in terms of both hypothetical (t [58] = 5.99, p = 0.001) and actual (z = 3.02, p = 0.003) rewards.

Conclusion

These data suggest that increased reactivity to environmental smoking cues contributes to the link between impulsivity and smoking.

Keywords

NicotineDisinhibitionReward

Introduction

Cigarette smoking is the most prevalent source of premature, preventable death and disease in the USA (Centers for Disease Control and Prevention 2002) and is associated with approximately 20% of all deaths (Centers of Disease and Control 1999). Despite the well-known deleterious effects of smoking, 23.3% of American adults remain regular smokers (Centers for Disease Control and Prevention 2002). The prevalence of smoking has decreased at a lower rate during the past 20 years relative to the previous two decades, and the cessation rates reported in clinical trials are significantly lower (Irvin and Brandon 2000; Irvin et al. 2003). One potential explanation for this change is that as the prevalence of smoking declines, those who continue to smoke may largely be those who have disproportionate difficulty quitting (Coambs et al. 1989; Hughes 1996), although this “hardening hypothesis” is controversial (e.g., Hughes and Brandon 2003; Warner and Burns 2003). It has been suggested that continuing smokers tend to possess underlying psychological and biological characteristics that are significant impediments to cessation (Gilbert and Gilbert 1995; Gilbert et al. 1998).

Previous research suggests that impulsivity, defined here as a tendency to pursue immediately rewarding environmental stimuli without regard for the consequences, is associated with smoking behavior. For example, smokers tend to be more impulsive than nonsmokers (Kassel et al. 1994; Lipkus et al. 1994; Zuckerman and Kuhlman 2000). Moreover, more impulsive smokers have increased difficulty quitting (Doran et al. 2004). However, relatively few studies have examined factors that might explain the association between impulsivity and cigarette smoking, and the precise nature of the relationship remains unclear (Mitchell 2004). The aim of the present study was to test the hypothesis that heightened impulsivity is associated with increased responsivity to environmental smoking cues.

Past research has shown that individuals who score highly on components of impulsivity such as lack of inhibitory control, lack of persistence, and sensation seeking are disproportionately drawn to reward cues and have great difficulty delaying or resisting a response to such cues (e.g., Buss and Plomin 1975; Jentsch and Taylor 1999). Thus, to the extent that cigarette smoking is rewarding, smoking-related cues may have a stronger “pull” for impulsive smokers than for others. Specifically, two reward-related processes may link impulsivity and smoking and may also explain impulsive smokers’ difficulty quitting. First, impulsive people are particularly aware of the presence of reward cues (Arnett and Newman 2000). For example, Nichols and Newman (1986) found that neurotic extraverts, who tend to be impulsive (Gray 1981; Pickering et al. 1999), responded more quickly than others to signals of reward availability on a computerized laboratory task. Impulsive individuals also tend to be either less aware of or less inhibited by punishment cues. For example, Corr et al. (1995) reported that in a two-stage laboratory learning task, impulsivity was associated with poor avoidance of a cued aversive stimulus.

A second key component of impulsivity is a preference for immediate rewards over larger, delayed rewards. This is often referred to as delay discounting, or the tendency to discount the value of delayed rewards relative to immediate rewards (e.g., Alessi and Petry 2003; Petry 2003). Interestingly, delay discounting may be particularly strong in the context of substance use, including smoking. Two studies found that smokers exhibited a greater preference for immediate cigarette rewards relative to immediate monetary rewards (Baker et al. 2003; Bickel et al. 1999). Delay discounting studies have generally assessed preferences for hypothetical rewards, and some researchers (e.g., Navarick 2004) have questioned the extent to which these findings can be generalized outside the laboratory. However, a number of studies comparing the discounting of hypothetical and actual rewards have failed to find an effect of reward type (e.g., Johnson and Bickel 2002; Lagorio and Madden 2005; Madden et al. 2003, 2004). While these findings suggest that hypothetical choice tasks are a reasonable approximation of an actual reward choice task, they have generally focused on monetary rewards. The present study is one of the first to compare the discounting of hypothetical and actual cigarette rewards. One potential weakness of the delay discounting literature in terms of smokers’ discounting of cigarette rewards is a lack of generalizability outside the laboratory. Smokers are likely to regularly encounter environmental stimuli that they associate with smoking, and research suggests that such stimuli are likely to be important determinants of smoking-related behavior (Carter and Tiffany 1999; Tiffany 1990). Examining discounting of cigarette rewards after exposure to a smoking cue may better approximate smokers’ experiences outside the laboratory.

In sum, more impulsive individuals are more likely than others to know when an immediate reward is present and obtainable and are less likely to inhibit behavior based on the presence of punishment cues. They are also more likely to seek to obtain a reward, with less regard for later negative consequences. In the context of smoking, more impulsive smokers may be especially likely to notice environmental cigarette cues (e.g., other smokers, advertisements, personal smoking triggers) and disproportionately prone to smoke when exposed to them. They may also find smoking cessation particularly difficult because it requires postponing gratification by choosing the delayed benefits of abstinence over the immediate gratification of smoking.

The purpose of the present study was to examine whether more impulsive smokers are more reactive than others to smoking cues and whether they have more difficulty than other smokers inhibiting immediate smoking after exposure to smoking cues. It was hypothesized that impulsivity would predict behavioral reactivity to smoking cues. That is, we expected that more impulsive smokers would have greater difficulty than other smokers delaying gratification after exposure to a smoking cue as compared to a neutral cue. We also expected that more impulsive smokers would exhibit a greater cigarette craving response to smoking cues than their less impulsive counterparts. Finally, this study was designed to compare impulsive smokers’ expected preference for immediate hypothetical and actual cigarette rewards.

Materials and methods

Sample

Participants (n = 60; 50% female) were regular smokers (i.e., at least 15 cigarettes per day for the past year) between the ages of 18 and 65 (M = 30.78, SD = 10.84). Thirty-three percent identified themselves as African-American, 9% as Asian-American, 51% as Caucasian, 4% as Latino-American, and 2% as multiethnic. Participants smoked an average of 19.11 (SD = 5.15) cigarettes a day, had smoked for an average of 18.07 years (SD = 9.90), and reported a mean score of 5.36 (SD = 1.88) on the Fagerstrom Test for Nicotine Dependence (FTND; Heatherton et al. 1991), suggesting a moderate level of nicotine dependence. Participants were recruited via flyers posted on campus at a large, public, Midwestern university and in the surrounding community and were paid to participate in one screening and two experimental sessions. Candidates currently meeting criteria for any Axis I disorder other than nicotine dependence were excluded.

Measures

Impulsivity Impulsivity was measured at the baseline screening session via self-report using the Barratt Impulsiveness Scale (BIS-11; Patton et al. 1995). The BIS-11 is a 30-item self-report measure that contains three subscales: motor impulsiveness (e.g., I am restless at the theater or lectures), attentional impulsiveness (e.g., I get easily bored when solving thought problems), and nonplanning impulsiveness (e.g., I spend or charge more than I earn). The BIS-11 has been shown to have good internal consistency and test–retest reliability and to consistently differentiate between impulsive populations (e.g., prisoners) and nonpsychiatric controls (Patton et al. 1995). Internal consistency in the present study was good (Cronbach’s α = 0.82), and total BIS-11 scores (M = 67.57) fell between those reported by Patton et al. 1995 for college students (M = 64.94) and several clinical samples (substance abuse patients, M = 69.00; psychiatric inpatients, M = 69.74; prison inmates, M = 76.30).

Nicotine dependence The FTND (Heatherton et al. 1991) was used to assess participants’ level of nicotine dependence. The FTND is a six-item self-report scale that sums behavioral responses suggestive of physiological dependence on nicotine, such as the time to first cigarette after wakening and smoking when ill. The FTND has demonstrated adequate internal consistency (Cronbach’s α = 0.64; Pomerleau et al. 1994) and strong test–retest reliability over a 3-week period (r = 0.88, p < 0.001, n = 60; Pomerleau et al. 1994). Internal consistency was acceptable in the present study (Cronbach’s α = 0.68). The FTND has also been found to be correlated with cotinine levels and years smoked (Pomerleau et al. 1994).

Cigarette craving Cigarette craving was measured via the Questionnaire of Smoking Urges—Brief (QSU-brief; Cox et al. 2001). The QSU-brief is a ten-item self-report measure of urge to smoke that yields two factor analytically derived subscales: cravings dealing with alleviation of negative affect and cravings that anticipate pleasure. The two subscales scores were added to create a single craving measure. Previous research indicates the QSU-brief has good construct validity and internal consistency (Cox et al. 2001). Internal consistency in the present sample was satisfactory, with Cronbach’s α coefficients ranging from 0.84 to 0.93.

Behavioral responsivity Behavioral cue responsivity was measured by participants’ choices in the cigarette choice task (CCT). Each subject completed the CCT twice, once after exposure to a smoking cue and once after exposure to a neutral cue. The CCT consisted of a hypothetical choice task and an actual choice task, which differed in terms of whether subjects received the cigarette reward that they chose. Subjects first completed the hypothetical choice task, which was administered in questionnaire format. Participants made a series of ten choices between one small, immediately available cigarette reward and one larger cigarette reward available (hypothetically) after a 1-h delay. In each of the ten choices, the immediately available reward was one cigarette, while the delayed reward varied. The delayed reward options were: 20, 15, 12, 10, 8, 6, 5, 4, 3, and 2 cigarettes, all available after a 1-h delay. The dependent measure was the break point or the point at which participants’ preference switched from the immediate to the delayed option. For example, if a participant preferred one cigarette immediately to three, four, five, six, and eight received after an hour but preferred 10, 12, 15, and 20 cigarettes after an hour to one immediately, the break point was coded as 10. Higher break point values indicated a stronger preference for the immediate cigarette reward.

The presentation of choices was counterbalanced, such that for half of the subjects, the delayed choices were presented in ascending order, while for the other half they were presented in descending order. After the hypothetical choice task, participants completed the actual choice task. They were told that they would spend an additional hour in the laboratory. Before the hour began, they were given a choice between one cigarette, which they would receive immediately, and eight cigarettes, which they would receive at the end of the hour. Subjects who selected the delayed reward were not permitted to smoke during the 1-h delay, whereas those who opted for the immediate reward were permitted to smoke only the one cigarette they received. For the actual choice task, the outcome measure was reward choice, a dichotomous variable (immediate vs delayed).

Smoking status At screening, candidates reported their daily cigarette consumption. At the beginning of each experimental session, participants were asked to report their smoking status and daily cigarette consumption since the last visit to the laboratory, as well as the amount of time elapsed since they last smoked. At screening and at the beginning of each experimental session, a noninvasive measure of exhaled carbon monoxide was used to measure smoking status. After inhaling deeply for 10 s, participants exhaled into a disposable cardboard tube placed over a sensor (EC-50, Vitalograph) to yield a measure of expired CO. Participants whose CO measured 8 ppm or higher at screening and at the beginning of each experimental session were considered to be smoking. Two of the 60 study participants had a CO value less than 8 ppm at baseline of an experimental session. In each case, the session was rescheduled, and the participant asked to smoke his/her normal amount before the new session; both had CO values greater than 8 ppm at baseline of the rescheduled session.

Procedure

Screening The initial screening was conducted via telephone. During the screening, candidates were told about the details of the study and given the opportunity to ask questions. Individuals who remained eligible and interested after the telephone screening were asked to come to the laboratory for an in-person screening interview (Structured Clinical Interview for DSM-IIIR, Nonpatient version; Spitzer et al. 1992).

Experimental sessions To approximate smokers’ experience in encountering smoking cues outside the laboratory, participants were instructed to smoke as usual before experimental sessions. Each 2-h experimental session began between 8 a.m. and 10 a.m. Upon arriving for experimental sessions, participants’ breath was analyzed for expired carbon monoxide to ensure that they had continued to smoke regularly after the screening session. They next spent approximately 20 min completing a series of questionnaires, including a measure of cigarette craving. After completing the questionnaires, participants were exposed to either a smoking cue or a neutral cue, as described below. Craving was reassessed after cue exposure, after which participants completed both the hypothetical CCT, followed by the actual CCT component. The two CCT components were administered in the same order (hypothetical first) in each experimental session to avoid the possibility that responses to the hypothetical CCT might be influenced by smoking for participants who opted for the immediate reward in the actual CCT.

Neutral vs smoking cue exposure conditions After completing the questionnaires, participants were exposed to either a neutral cue or a smoking cue. The cue exposure procedure was adapted from the procedure described by Sayette and Hufford (1994). In the smoking cue condition, participants were shown a pack of their usual brand of cigarettes, a lighter, and an ashtray on the table in front of them. They were instructed to light a cigarette without touching the cigarette to their mouths. Participants were next instructed to hold the cigarette in their hand without smoking it. They sat with the cigarette in their hand for 5 min. The control cue procedure was similar to the smoking cue procedure. Rather than a cigarette, participants were given a roll of tape to hold in their hand for 5 min.

Analytic plan

It was hypothesized that more impulsive smokers would exhibit disproportionate reactivity to smoking cue exposure as compared to neutral cue exposure. Cue reactivity was measured by changes in cigarette craving and greater preference for immediate cigarette rewards vs larger, delayed cigarette rewards after exposure to a smoking cue compared to a neutral cue. Changes in craving were assessed using mixed effects regression to model change over time (baseline to postexposure) and condition (cigarette vs neutral cue), implemented via SAS PROC MIXED. Impulsivity was included as a predictor variable in the model, and a significant impulsivity × condition × time interaction was considered to support the hypothesis. A significant interaction was followed by stratifying the data by cue condition and testing the impulsivity × time interaction within each condition. It is important to note that the inclusion of interaction terms alters the interpretation of main effects. Rather than indicating the overall effect of a variable, main effects reflect the effect of the variable when the interacting variables equal 0. For example, the model for craving includes the three-way interaction between impulsivity, condition, and time. Consequently, the estimate for the main effect of impulsivity reflects the effect when both condition and time equal 0 (i.e., baseline craving assessment in the neutral cue condition).

Impulsivity- and condition-related differences in reward preference were examined by testing the effects of impulsivity and condition on break point (in the hypothetical CCT) and actual choice (in the actual CCT). Mixed-effects regression was used to test the effects of impulsivity and condition on break point. Analysis of the effects of impulsivity and condition on actual choice was performed using a logistic regression model for correlated dichotomous outcomes estimated via the generalized estimating equations method (Liang and Zeger 1986). For both analyses, support for the hypothesis consisted of a significant impulsivity × condition interaction. Significant interactions were followed by stratifying the data by cue condition and using linear or logistic regression to test the effect of impulsivity within each condition.

Longitudinal regression modeling techniques were chosen rather than univariate (e.g., analysis of variance) techniques for two primary reasons. First, these techniques permit the inclusion of time-varying (e.g., cigarette craving) and time-invariant (e.g., impulsivity) variables, as well as both continuous and categorical variables, in a single model. Consequently, these analyses allow assessment of effects that are stable as well as those that change over time. Second, these methods are likely to yield more accurate models than univariate techniques, which estimate average change over time, as it is unlikely that the rate of change is the same across participants (Hedeker 2002). The longitudinal modeling techniques used in this study estimate change for each individual rather than averaging across individuals and therefore provide a more accurate representation of the data.

Results

Preliminary analyses

All continuous predictor and outcome variables (impulsivity, craving over time, break point) were found to be normally distributed. As recommended by Keppel (1991), variables that were significantly associated with either impulsivity or one or more outcome measures were included as covariates. Age was found to be significantly correlated with impulsivity (r = −0.38, p = 0.004) and was therefore included as a covariate in all analyses. Because previous research has shown that men tend to be more impulsive than women (e.g., Waldeck and Miller 1997), gender was also included as a covariate. Variables that were found to be independent of all predictor and outcome variables and were excluded from analyses included household income, education, marital status, order of cue exposure sessions (i.e., neutral first vs cigarette first), exposure session sequence (i.e., first session vs second session), and nicotine dependence.

Because participants were instructed to smoke as usual before experimental sessions, we considered the possibility that responses could have been influenced by nicotine deprivation. Correlational analyses indicated that time since last cigarette was not significantly associated with impulsivity, baseline craving at either experimental session, changes in craving after cue exposure, or break point in the hypothetical CCT. Logistic regression showed that time since last cigarette was also independent of reward choice on the actual CCT. To further address this issue, we tested the relationship of baseline CO to impulsivity and all outcome measures. Baseline CO values were not significantly correlated with impulsivity or with post-exposure change in craving, regardless of whether baseline craving scores were included as a covariate. Similarly, CO was not associated with either break point or actual reward choice in the CCT.

Manipulation check To determine the effect of the manipulation, we compared cigarette craving and reward choice between the two conditions. Means and standard deviations for each variable are shown in Table 1. Craving increased significantly from baseline to post-exposure in both the neutral (t [59] = 6.90, p = 0.001) and cigarette cue (t [59] = 6.04, p = 0.001) conditions. Post-exposure craving was significantly higher in the cigarette cue condition than in the neutral cue condition (t [59] = 2.10, p = 0.040). Additionally, in both the hypothetical (t [59] = 2.63, p = 0.011) and actual (χ2 [1] = 6.85, p = 0.009) components of the CCT, participants exhibited a greater preference for immediate rewards in the cigarette condition compared to the neutral condition.
Table 1

Descriptive statistics and cross-condition comparisons of outcome variables

Variable

Neutral cue condition M (SD)

Cigarette cue condition M (SD)

Cigarette craving

Baseline

33.90 (10.82)

33.77 (10.91)

Postexposurea

37.51 (10.92)b

42.52 (14.00)b

Hypothetical reward choicea

5.82 (2.70)

6.71 (3.86)

Actual reward choicea

30.6%

59.7%

Hypothetical reward choice values indicate break point or the point at which participants’ reward preference changed from larger, delayed rewards to smaller, immediate rewards; higher values indicate greater preference for immediate rewards. Actual reward choice values indicate percent choosing immediate reward.

aSignificant difference between conditions (p < 0.05)

bSignificant difference from baseline to postexposure (p < 0.05)

Primary analyses

Cigarette craving Mixed effects regression models were employed to test the effect of impulsivity on change in craving after cue exposure. The dependent variable in each model consisted of the four craving measurements over time (i.e., pre- and post-exposure in the cigarette cue and neutral cue conditions) to account for baseline levels of craving. As Table 2 shows, age and gender were unrelated to total craving scores. The significant time × condition interaction (t [161] = −2.57, p = 0.011) indicates a larger post-exposure increase in craving in the cigarette cue condition than in the neutral cue condition. As predicted, the impulsivity × condition × time interaction was significant (t [161] = 3.21, p = 0.002). To interpret the interaction, the data were stratified by condition, and the effect of impulsivity on craving over time was tested within each condition. These analyses revealed significant impulsivity × time interactions in both the neutral (t [107] = 3.00, p = 0.004) and cigarette cue (t [107] = 7.15, p < 0.001) conditions, meaning that more impulsive participants reported greater increases in craving than others in both conditions. Because impulsivity and craving were measured continuously, we plotted correlations between the two by time and condition to depict the relationship graphically. As shown in Fig. 1, impulsivity was significantly associated with craving at each measurement; the relationship was strongest after exposure to the cigarette cue.
Table 2

Effect of impulsivity on change in QSU-brief total scores over time

Variable

Regression coefficienta

Standard error

t statistic

Covariates

Time

−5.88

6.11

−0.96

Age

−0.10

0.31

−0.32

Gender

−2.73

2.48

−1.10

Predictors

Exposure condition

0.39

6.11

0.06

Time × condition

−22.21

8.64

−2.57**

Impulsivity

0.14

0.12

1.15

Impulsivity × condition

−0.01

0.09

−0.14

Impulsivity × time

0.14

0.09

1.58

Impulsivity × condition × time

0.41

0.13

3.21*

*p < 0.01

**p < 0.05

aUnstandardized

https://static-content.springer.com/image/art%3A10.1007%2Fs00213-007-0832-x/MediaObjects/213_2007_832_Fig1_HTML.gif
Fig. 1

Correlations between impulsivity and cigarette craving over time by cue exposure condition

To further specify between-condition differences, we created craving difference scores by subtracting baseline craving from postexposure craving scores. We then calculated partial correlation coefficients between impulsivity and craving difference scores, adjusted for baseline craving, for each cue exposure condition separately. These analyses yielded significant correlations in both the cigarette cue condition (r = 0.42, p = 0.001) and the neutral cue condition (r = 0.28, p = 0.032). However, this effect was disproportionately strong in the cigarette cue condition. In other words, more impulsive smokers reported greater increases in craving when exposed to either cue but particularly when exposed to the smoking cue.

Cigarette choice task The mixed effects regression model of the hypothetical choice task is shown in Table 3. Age and gender were unrelated to break point on the hypothetical choice task. The main effect of impulsivity was significant (t [59] = 6.18, p < 0.001). There was a significant impulsivity × condition effect (t [59] = 5.99, p < 0.001) on break point. In other words, more impulsive participants’ break point was higher than others’ was, across exposure conditions, indicating a greater preference for immediate rewards. This effect was stronger after cigarette cue exposure than after neutral cue exposure.
Table 3

Effect of impulsivity on preference for hypothetical immediate cigarette rewards after cigarette cue compared to neutral cue exposure

Variable

Regression coefficienta

Standard error

t statistic

Covariates

Age

0.26

0.35

0.74

Gender

0.19

0.59

0.31

Predictors

Exposure condition

−8.54

1.57

−5.45*

Impulsivity

0.14

0.02

6.18*

Impulsivity × condition

0.13

0.02

5.99*

*p < 0.01

aUnstandardized

Logistic regression models for correlated dichotomous outcomes were used to assess the effect of impulsivity on the actual choice component of the CCT and yielded results consistent with analyses of the hypothetical component of the task (Table 4). There were significant main effects of cue exposure condition (z = −2.55, p = 0.011) and gender (z = 3.93, p < 0.001), indicating that participants were more likely to choose the immediate reward in the cigarette cue condition than in the neutral cue condition and that men were more likely than women to opt for the immediate reward regardless of condition. There was a main effect of impulsivity (z = −4.04, p < 0.001), such that those with high levels of impulsivity were more likely to select the immediate reward after exposure to a neutral cue. The impulsivity × condition interaction (z = 3.02, p = 0.003) indicates a similar but stronger effect after cigarette cue exposure. In other words, higher levels of impulsivity were associated with a greater probability of choosing the immediate reward in both conditions but particularly so in the cigarette cue condition.
Table 4

Effect of impulsivity on preference for actual immediate cigarette rewards after cigarette cue compared to neutral cue exposure

Variable

Regression coefficienta

Standard error

z statistic

Covariates

Age

−0.51

0.49

−1.03

Gender

11.83

3.01

3.93*

Predictors

Exposure condition

−7.47

2.93

−2.55**

Impulsivity

−0.18

0.05

−4.04*

Impulsivity × condition

0.13

0.04

3.02*

*p < 0.01

**p < 0.05

aUnstandardized

Finally, we used logistic regression to examine agreement between the two components of the CCT. Break point in the hypothetical component was found to be a significant predictor of reward choice in the actual component in both the cigarette cue (odds ratio [OR] = 1.37 [95% confidence interval {CI} 1.12–1.67], p = 0.002) and neutral cue (OR = 1.39 [95% CI 1.14–1.70], p = 0.001). That is, across cue exposure conditions, participants with higher (i.e., more impulsive) break point values were more likely to choose the immediate actual reward, suggesting generally consistent responding across the two tasks.

Discussion

The current study tested the hypothesis that more impulsive smokers would exhibit stronger reactions to cigarette cues than other smokers. We found that impulsivity was associated with a disproportionate cigarette craving response to both cigarette cue exposure and neutral cue exposure but was most strongly associated with increased craving after exposure to a cigarette cue. Impulsivity was also found to be associated with a greater preference for immediately available cigarette rewards rather than larger, delayed cigarette rewards in both a hypothetical choice task and an “actual” choice task.

These findings suggest two candidate mechanisms that could contribute to an explanation of the link between smoking and impulsivity. First, more impulsive smokers reported disproportionate levels of cigarette craving in general and particularly in response to a smoking cue. As craving is typically considered an important component of nicotine dependence (e.g., Shiffman et al. 2004), this suggests that more impulsive smokers are likely to be more dependent on nicotine compared with other smokers. In the community, there are likely to be countless environmental smoking cues (e.g., others smoking, advertisements, individual triggers). More impulsive smokers appear likely to experience stronger cravings than others in response to smoking cues. Additionally, two recent studies indicate that an elevated craving response to smoking cues is associated with greater difficulty quitting smoking (Tiffany et al. 2000; Waters et al. 2004). Second, after exposure to a cigarette cue, more impulsive smokers had greater difficulty inhibiting immediate smoking despite the fact that doing so would have maximized overall reward. This finding is also consistent with higher levels of nicotine dependence and greater cessation difficulty. It may be that more impulsive smokers tend to prefer the immediate outcomes associated with continuing to smoke over the delayed outcomes associated with quitting to a greater extent than less impulsive smokers. The extent to which these two candidate mechanisms are separate is unclear. It is plausible that more impulsive smokers’ preference for immediate smoking after cigarette cue exposure is mediated by a stronger craving response or that both are associated with a third variable (e.g., deficient dopaminergic function). Additional research is needed to determine the extent to which impulsive smokers’ difficulty delaying smoking is explained by differential cigarette craving.

The present findings add to the scientific literature on the discounting of delayed drug rewards. Several previous studies have shown that smokers tend to discount the value of delayed monetary (Bickel et al. 1999; Mitchell 1999), health (Baker et al. 2003), and cigarette (Bickel et al. 1999) outcomes to a greater extent than nonsmokers on hypothetical choice tasks. These findings indicate that smokers tend to be more impulsive (i.e., have greater difficulty delaying gratification) than nonsmokers. Most studies of delay discounting have utilized hypothetical rewards. Although this is more convenient for research purposes (Baker et al. 2003), researchers have recognized the possibility that subjects respond differently when choosing between outcomes that they know will not occur. Other studies (e.g., Crean et al. 2000; Mitchell 1999) have used “potentially real” rewards, in which subjects make a number of choices and know that they will receive one of the rewards they choose. Finally, a few recent studies have directly compared discounting in the hypothetical and potentially real methodologies (e.g., Lagorio and Madden 2005; Madden et al. 2004). Although the two methodologies have generally produced similar results, there has continued to be concern that subjects might be responding to a situation that is not representative of the world outside the laboratory (Bickel and Marsch 2001; Madden et al. 2004; Navarick 2004). To our knowledge, the present study was one of the first to use a procedure in which subjects received all rewards they selected in a particular task and the first to do so with cigarette rewards. Additionally, the inclusion of a hypothetical choice task allows for comparison of the two methods. Notably, we found similar results for both procedures, suggesting that hypothetical choice tasks reasonably approximate actual choice.

Certain aspects of the present study may limit its generalizability. In particular, the forced choice aspect of the actual component CCT created an artificial situation that may not be representative of smokers’ experiences. When choosing between the two cigarette rewards, participants were aware that regardless of their choice, 1 h later they would be free to smoke as many of their own cigarettes as they desired. This knowledge may have led to increased preference for the immediate reward. Participants may have had the perception that by selecting one cigarette immediately, they received both rewards, possibly influencing the measurement of their tendency to discount delayed rewards. This situation is somewhat analogous to circumstances outside the laboratory in which smokers may be required to delay smoking (e.g., in the workplace). As such, the present procedure may be a valid method for assessing the ability to inhibit immediate smoking after exposure to smoking cues. Additionally, it should be noted that in the present study, the hypothetical procedure was always completed immediately before the actual procedure. It is possible that participants’ responses on the hypothetical choice task created a demand for them to respond similarly on the actual CCT. It is also possible that the generalizability of the present study may be limited by the criteria used to determine participant eligibility. Specifically, candidates currently meeting criteria for Axis I disorders other than nicotine dependence were excluded. Given that high levels of impulsivity tend to be comorbid with a variety of such disorders, particularly substance use disorders, it may be that the current sample was not representative of highly impulsive smokers in the community. Caution should be exercised in generalizing these findings to such individuals.

Another potential limitation of the findings is that the cue-induced change in craving was fairly small. As shown in Table 1, mean craving ratings increased by approximately nine points after smoking cue exposure and four points after neutral cue exposure. The size of the difference between cue conditions in postexposure craving (Cohen’s d = 0.41) was considerably less than the mean effect size of 1.18 reported in a recent meta-analysis (Carter and Tiffany 1999), suggesting somewhat limited potency of the cue manipulation. Some previous studies have reported effects of similar magnitude among nondeprived smokers (e.g., Rickard-Figueroa and Zeichner 1985; Sayette and Hufford 1995), and increased deprivation has been shown in some studies to predict greater increases in craving after smoking cue exposure (e.g., Payne et al. 1996). Thus, one possible explanation for the relatively small effect is that participants were not asked to refrain from smoking before experimental sessions. Participants’ perceived availability of smoking may also have contributed to the relatively small effect. Craving responses to smoking cues have been shown to be stronger when participants believe they will be able to smoke shortly (Carter and Tiffany 2001; Dols et al. 2002; Juliano and Brandon 1998). In the current study, participants were informed that they would remain in the laboratory for an hour after cue exposure and were not necessarily aware that they would have the opportunity to smoke immediately, potentially limiting post-exposure changes in craving. Despite the relatively small effect on craving, smoking cue exposure significantly increased preference for immediate cigarette rewards, particularly among more impulsive smokers. To the extent that these data underestimate the effect of smoking cues on cigarette craving, they may also underestimate the effect of smoking cues on preference for smoking immediately.

In sum, our findings suggest that more impulsive smokers are likely to experience disproportionately strong cigarette cravings when exposed to an environmental stimulus that reminds them of smoking. Impulsive smokers exhibited a stronger predilection for immediate compared to delayed rewards than others; this discrepancy was exacerbated after exposure to a smoking cue. Reward choices were similar for hypothetical and actual rewards, supporting the validity of hypothetical reward choice tasks. These findings suggest that more impulsive smokers, when exposed to a smoking cue, tend to develop disproportionately strong cigarette cravings. Perhaps, this heightened craving response gives more impulsive smokers disproportionate difficulty inhibiting immediate smoking after being exposed to a smoking cue.

Although the present findings add to knowledge about mechanisms that may link impulsivity and cigarette smoking, further research is needed to fully explain the relationship. For example, future research should test whether impulsive smokers’ increased responsivity to smoking cues is associated with greater difficulty quitting smoking. To the extent that responsivity to smoking cues does represent a pathway to relapse among more impulsive smokers, it may also be useful to assess whether teaching such individuals alternative skills for responding to smoking cues might attenuate this effect. Additionally, as Mitchell (2004) has noted, it is not presently clear whether impulsivity is a risk factor for smoking initiation and the development of dependence or whether smoking leads to increased impulsivity. Evidence to directly address this question is sparse but suggests that both models may be accurate. In light of the fact that the answer likely has important implications for our understanding of the role of impulsivity in smoking, as well as for prevention and cessation efforts, further systematic research is needed.

Acknowledgment

This research was supported by the American Heart Association grant 0410025Z to Neal Doran.

Copyright information

© Springer-Verlag 2007