Psychopharmacology (1999) 146:447–454
© Springer-Verlag 1999
O R I G I N A L I N V E S T I G AT I O N
Warren K. Bickel · Amy L. Odum Gregory J. Madden
Impulsivity and cigarette smoking: delay discounting in current, never, and ex-smokers
Received: 3 March 1999 / Final version: 11 May 1999
Abstract Rationale: Impulsivity is implicated in drug dependence. Recent studies show problems with alcohol and opioid dependence are associated with rapid discounting of the value of delayed outcomes. Furthermore, discounting may be particularly steep for the drug of dependence. Objectives: We determined if these findings could be extended to the behavior of cigarette smokers. In particular, we compared the discounting of hypothetical monetary outcomes by current, never, and ex-smokers of cigarettes. We also examined discounting of delayed hypothetical cigarettes by current smokers. Methods: Current cigarette smokers (n=23), never-smokers (n=22) and ex-smokers (n=21) indicated preference for immediate versus delayed money in a titration procedure that determined indifference points at various delays. The titration procedure was repeated with cigarettes for smokers. The degree to which the delayed outcomes were discounted was estimated with two non-linear decay models: an exponential model and a hyperbolic model. Results: Current smokers discounted the value of delayed money more than did the comparison groups. Never- and ex-smokers did not differ in their discounting of money. For current smokers, delayed cigarettes lost subjective value more rapidly than delayed money. The hyperbolic equation provided better fits to the data than did the exponential equation for 74 out of 89 comparisons. Conclusions: Cigarette smoking, like other forms of drug dependence, is characterized by rapid loss of subjective value for delayed outcomes, particularly for the drug of dependence. Never- and ex-smokers could discount similarly because cigarette smoking is associated with a reversible increase in discounting or due to selection bias. Key words Delay discounting · Impulsivity · Drug abuse · Cigarette · Nicotine W.K. Bickel (✉) · A.L. Odum · G.J. Madden Human Behavioral Pharmacology Laboratory, Department of Psychiatry, 38 Fletcher Place, University of Vermont, Burlington, VT 05401, USA e-mail:
[email protected], Fax: +1-802-656-9628
Introduction Impulsive behavior is frequently observed in drug dependence. For example, drug-dependent individuals, as indicated by DSM-IIIR diagnostic criteria, often forgo or reduce occupational or recreational activities to use drugs. Drug-dependent individuals select relatively brief but immediately available bouts of drug intoxication or relief of transient withdrawal symptoms over a variety of prosocial, but often deferred, rewards. Similarly, intravenous drug users, in the pursuit of immediate rewards, may opt to share hypodermic needles instead of deferring drug use until they have time to disinfect or obtain clean needles. Perhaps understanding of drug dependence may be improved by the scientific investigation of this impulsive behavior. However, other than noting that drug-dependent individuals exhibit what may be deemed impulsive behavior (Logue 1995), and that they score higher than controls on standardized measures of impulsivity (e.g., Alcock and Grace 1988; Rosenthal et al. 1990), research has not yet rigorously identified the prevalence or the degree of impulsivity that may be evident in different types of drug-dependence disorders. Moreover, drug-dependent individuals may not be equally impulsive on all choices. Impulsivity could vary with the type of reward (see Raineri and Rachlin 1993; Madden et al. 1997). Impulsivity, in addition to being viewed as a personality characteristic (e.g., Eysenck and Eysenck 1978; Webster and Jackson 1997), may also reveal an important aspect of behavior: the effect on behavior of reinforcer amount is modulated by the delay to the delivery of that reinforcer (Logue 1988). Said another way, the value of a delayed reward is discounted (reduced in value or considered to be worth less) compared to the value of an immediate reward. One account of delay discounting has been provided by economics (e.g., Lancaster 1963; Meyer 1976; Fishburn and Rubinstein 1982). Economic models, including the theory of rational addiction (Becker and Murphy 1988), assume that delayed rewards are dis-
448
counted exponentially: for each unit of time that constitutes the delay to delivery, the value of a reward decreases (or is discounted) by a fixed proportion. An exponential function commonly used to describe the effect of delay on reward value is: V=Ae–kD
(1)
where V is the subjective value of a reward, A is the amount of the reward, and D is the delay to the reward. The free parameter k is a measure of discounting rate as it determines how rapidly the function decreases: the larger k, the more rapidly the subjective value of outcomes declines with increasing delays. However, exponential discounting has not been empirically supported by behavioral research conducted in non-human and human subjects. Instead, these studies demonstrate that delay discounting is hyperbolic. Hyperbolic discounting refers to the devaluation of delayed rewards proportional to their delay (Ainslie 1992). For each unit of time that constitutes the delay to delivery, the reward’s present value decreases by an increasingly smaller proportion. Studies examining delay discounting in human subjects typically employ procedures similar to those used in psychophysical experiments (Richards et al. 1997). Psychophysical procedures usually present a participant with a standard stimulus and then with a stimulus that is adjusted until the two stimuli are considered equivalent by the participant (Stevens 1975). Similarly, procedures used in delay discounting experiments present subjects with a choice between a standard larger-later reward (e.g., $1000 delivered in 1 year) and an immediate reward the magnitude of which is adjusted until the subjective value of the two rewards are approximately equal (e.g., Rachlin et al. 1991). This point of equivalence is the indifference point for that particular delay interval. When indifference points are obtained for a variety of delays, an indifference curve may be plotted. Indifference curves permit the shape of the function to be empirically determined and the degree to which delayed rewards are discounted to be empirically derived (see Mazur 1987). The degree of discounting can be calculated by application of the following hyperbolic equation developed by Mazur (1987): V=A/1+kD.
(2)
In Equation 2, V is the present discounted value of a delayed reward (i.e., the indifference point), A is the amount of the delayed reward, k is an empirically derived constant proportional to the degree of delay discounting, and D is delay duration. Empirically determined indifference curves have been demonstrated to be hyperbolic and Equation 2 has been found to provide an accurate model of the discounting function when food and water are used with non-humans (e.g., Mazur 1987; Rodriguez and Logue 1988; Richards et al. 1997) and real and hypothetical money are used with human subjects (e.g., Rachlin et al. 1991; Green et al. 1994;
Myerson and Green 1995; Kirby 1997; Madden et al. 1997; Kirby et al. 1999). Although delay discounting appears related to issues of drug dependence, it has rarely been studied in drug-dependent individuals. Until recently (Mitchell 1999), prior studies had examined discounting functions only with participants with alcohol (Vuchinich and Simpson 1998) or heroin (e.g., Madden et al. 1997) use problems. Vuchinich and Simpson (1998) examined delay discounting of hypothetical monetary rewards in participants who were heavy versus light drinkers of alcohol and problem versus light drinkers of alcohol. The participants were college students not in treatment. Relative to the light drinkers, heavy and problem drinkers showed more extreme discounting (i.e., heavy and problem drinkers devalued delayed rewards more than did light drinkers). Moreover, the hyperbolic equation described above accounted for significantly more variance than the exponential function developed via economic approaches. We have also conducted studies using similar procedures with opioid-dependent outpatients. Our initial study compared the delay discounting of opioid-dependent outpatients in treatment and matched control participants (Madden et al. 1997). Participants chose between hypothetical monetary rewards available immediately or following a delay. The opioid-dependent participants were also given a second series of choices between hypothetical immediate and delayed heroin using the same procedures (the amount of heroin was adjusted using street values such that the dollar amount of the delayed heroin was $1000). Opioid-dependent outpatients discounted money to a significantly greater degree than did control participants. Within the opioid-dependent sample, the discounting of heroin was significantly greater than the discounting of money, demonstrating that the magnitude of discounting may not be invariant, but may depend upon the type of rewards available. In another study, we systematically replicated the preceding one using procedures in which actual monetary outcomes were delivered for a subset of choices (Kirby et al. 1999). Opioid-dependent outpatients were found to discount delayed money to a significantly greater degree than did control participants, confirming that the same overall finding observed with hypothetical rewards could be replicated with actual monetary outcomes. An additional study systematically replicated the first study with two important additions: (1) hyperbolic versus exponential curves were compared to determine which better fit the data and (2) the amount of heroin that was equivalent to $1000 was adjusted according to the value for each subject (Madden et al. 1999). The study demonstrated that the obtained discounting functions were better accounted for by the hyperbolic form of the discounting equation (Equation 2). Furthermore, a recent study examined the hypothetical monetary discounting of opioid-dependent participants who reported sharing intravenous needles or did not do so (Odum et al., unpublished data). Money was discounted significantly more by delay in the sharing group than in the non-sharing group.
449
The purpose of the present study was to expand our observations to a new drug dependent group: cigarette smokers. Cigarettes smokers have been shown to demonstrate greater impulsivity than non-smokers on personality measures (e.g., Geist and Herrmann 1990; Lipkus et al. 1994; Waldeck and Miller 1997); only recently has research examined whether smokers discount future events more so than non-smokers (Mitchell 1999). Discounting by cigarette smokers may be important to examine because smokers typically do not experience major disruptions in their life status (e.g., homelessness, loss of employment) as a result of their dependency, unlike those who are dependent upon heroin or alcohol (cf. Hughes 1993). The extreme discounting of future outcomes shown by previously studied groups with drug problems could be in part generated indirectly by the disruptions in life status these groups experience. Specifically, the objectives of this study were 3-fold. First, we compared the extent to which delayed monetary rewards were discounted in value by cigarette smokers, a matched group of never-smokers, and a matched group of ex-smokers. This comparison may permit an understanding of whether greater discounting is an enduring feature of an individual, similar to personality accounts, or whether discounting is related to being currently nicotine dependent. Second, we compared cigarette smokers’ discounting of delayed monetary rewards with their discounting of an equivalent amount of cigarettes to determine whether the greater discounting of the drug of dependence than for money seen in heroin-dependent participants was also observed in cigarette smokers. Third, we compared exponential and hyperbolic equations to determine which better fit the discounting data.
Materials and methods Participants Participants were 23 current cigarette smokers, 21 ex-smokers, and 22 never-smokers recruited through newspaper advertisements and by fliers posted in prominent public places. Participants were screened for substance abuse and psychiatric disorders (the exception being nicotine dependence for current cigarette smokers). Current smokers were those who reported smoking at least 20 cigarettes per day for at least the past 5 years and had a Fagerström Test for Nicotine Dependence (FTND) score of at least 6. Neversmokers were those who reported never smoking and whose demographic characteristics (shown in Table 1) matched those of participants in the current smoker group. Ex-smokers were those who reported abstinence from cigarettes for at least one year, had smoked at least 20 cigarettes per day for at least 5 years prior to Table 1 Demographic and participant characteristics and Eysenck Impulsivity score for current, never, and ex-smokers Unless indicated otherwise, values listed are group means (SD) a Median (interquartile range)
quitting, and had demographic characteristics (shown in Table 1) that matched those of participants in the current smoker group. Participants volunteered and gave written informed consent for the study which was approved by the University of Vermont Institutional Review Board. Participants were tested individually and received either $35 (smokers) or $25 (ex-smokers and never-smokers) compensation for an approximately 2-h session. Smokers received greater compensation because they completed an additional procedure involving cigarettes (described below). Materials and procedure Participants were seated individually in a small quiet room across a table from the research assistant. They provided demographic information and completed the Eysenck Personality Questionnaire (Eysenck and Eysenck 1978) and other scales and tasks (not analyzed here). The research assistant administered the information, arithmetic, picture completion, and block design tests of the Wechsler Adult Intelligence Scale – Revised (WAIS-R) to provide an estimate of IQ, then conducted the delay discounting assessment. Never-smokers and ex-smokers received the monetary version of the assessment. Smokers received in addition the cigarette version of the assessment. The order of the assessments (monetary and cigarette) was counterbalanced across smokers. Because cigarette abstinence could plausibly affect performance on these tasks, current smokers were instructed to smoke as usual before they arrived for testing and took a scheduled cigarette break after the first assessment. For the delay discounting assessment, monetary rewards, as well as the delays to the rewards, were printed on separate 3×5 index cards. The 27 monetary reward amounts were $1000, $990, $960, $920, $850, $800, $750, $700, $650, $600, $550, $500, $450, $400, $350, $300, $250, $200, $150, $100, $80, $60, $40, $20, $10, $5, and $1. There were two $1000 cards. The seven reward delays were 1 week, 2 weeks, 1 month, 6 months, 1 year, 5 years, and 25 years. These amounts and delays have commonly been used in the study of delay discounting (e.g., Rachlin et al. 1991). Current smokers were also tested with cigarettes. To ensure the magnitude of the monetary and cigarette rewards was equal, the current smokers were asked how many cartons of cigarettes they could purchase with $1000. The research assistant helped as necessary and verified the calculation based on the price per carton or pack the participant stated. This number was then multiplied by the following values to produce cards of equivalent amounts to the monetary rewards: 1.0, 0.99, 0.96, 0.92, 0.85, 0.80, 0.75, 0.70, 0.65, 0.55, 0.50, 0.45, 0.40, 0.35, 0.30, 0.25, 0.20, 0.15, 0.10, 0.08, 0.06, 0.04, 0.02, 0.01, 0.005, and 0.001. Amounts were expressed in the number of cartons, and in the case of fractions of cartons, number of packs and/or individual cigarettes as necessary. These figures were printed on 28 separate 3×5 index cards (there were two cards for the highest amount) prepared for each current smoker participant. Delays were the same as for the monetary task. Participants chose between amounts of money (or cigarettes for the cigarette task) delivered immediately and an amount delivered after a delay. The index cards describing the amounts and delays were placed on the table facing the seated participant. To the participant’s left was a stack of cards (above which was a card reading “Right Now”) indicating amounts that could be received immediately, and to the participant’s right a single card, labeled
Variable
Current (n=23)
Never (n=22)
Ex (n=21)
Sex (% male) Age (years) Education (years) Income ($ monthly)a IQ Eysenck Impulsivity
48 33.7 (8.8) 14.3 (1.9) 800 (375–1550) 102.1 (11.0) 10.4 (4.9)
32 32.6 (9.1) 14.9 (1.6) 950 (225–1500) 102.7 (9.5) 9.1 (4.1)
38 35.7 (9.5) 14.7 (1.6) 800 (0–1000) 103.8 (10.8) 7.5 (3.4)
450 $1000, which could be received after some delay (the duration of which was indicated on a card above). The research assistant turned through the 27 immediate cards, starting with $1000, down to $1, and back to $1000. For each card, the participant indicated whether he or she preferred the immediate or delayed reward. The research assistant noted the last immediate reward chosen in the descending order, and the first immediate reward chosen in the ascending order, and the average of these was taken as the point of subjective equality (i.e., the point at which the immediate and delayed rewards were equal in value). This procedure was repeated at each of the seven delays. Instructions The research assistant read the following instructions to all participants prior to beginning the delay discounting assessment: Now I am going to ask you to make some decisions about which of two consequences you prefer. You will not receive the consequences that you choose, but we want you to make your decisions as though you were really going to get the consequences you choose. The choices you make are completely up to you. Please select the option that you prefer, not what you might think I might want you to prefer. I do not expect you to choose one particular consequence over another. Just choose the consequence that you really want. For the monetary reward task (all groups), the research assistant read the following additional instructions: Now I want you to make some choices involving money. The card on your left shows some money that you can have right now. The card on the right shows money that you can have after you have waited some period of time. So now you are being asked to choose between $1000 delivered today versus $1000 that you would get 1 week from today. Point to the reward that you would prefer. [Participant points to preferred reward, in all cases $1000 delivered today]. Now I’m going to keep flipping through these cards. Keep pointing to the reward that you would prefer. For cigarette smokers, the procedure was also conducted with cigarettes after the research assistant read the following additional instructions: In the next set of consequences, you will be asked to choose between two different cigarette outcomes. I would like you to make your decisions as though you were really going to get the consequences you choose. The choices you make are completely up to you. Please select the option you prefer, not what you think I might want you to prefer. I do not expect you to choose one particular consequence over another. Just choose the consequence that you really want. The card on your left shows a cigarette credit that you can have right now at your favorite store. That is, the store will allow you to have cigarettes for free, starting today, until you use up your credit. The card on the right shows a cigarette credit that you can have also at your favorite store, after you have waited for some period of time. So now you are being asked to choose between [the number of cartons that participant could purchase with $1000] cartons of cigarettes that you can start using today versus [the number of cartons that participant could purchase with $1000] cartons of cigarettes that you would get 1 week from today. Point to the reward that you would prefer. [Participant points to preferred reward, in all cases $1000 worth of cigarettes available starting today]. Keep pointing to the reward that you would prefer as I flip through these cards. Analyses Normally distributed data for demographic and participant characteristics and Eysenck Impulsivity Score were assessed with a oneway analysis of variance (ANOVA). Planned pairwise compari-
sons for Eysenck Impulsivity Score for current smokers versus never-smokers, current smokers versus ex-smokers, and neversmokers versus ex-smokers were conducted by two-tailed t-tests. Income was not normally distributed and was therefore analyzed with the Kruskal-Wallis one-way ANOVA. The male-female proportion in each group was compared by a Chi-square test. For the delay discounting assessment, the hyperbolic decay and exponential models were fit to data from each participant for the money condition and to data from current smokers for the cigarette condition using non-linear regression (GraphPad Prism®). When the models are fit to the data, the resulting distributions of the derived discounting parameter (k ) are often non-normal and thus evaluated with non-parametric statistics (Rachlin et al. 1991). The Kruskal-Wallis one-way ANOVA was used to examine the degree of discounting (k) estimated by the hyperbolic equation for money for current smokers, never-smokers, and ex-smokers. Planned pairwise comparisons of the discounting parameter (k) estimated by the hyperbolic equation for money between current smokers and never-smokers, current smokers and ex-smokers, and never-smokers and ex-smokers were conducted with the MannWhitney U-test. The Wilcoxon signed-ranks test was used for within-group comparisons of the discounting parameter estimated by the hyperbolic equation for current smokers for money versus cigarettes. To evaluate the fits for the hyperbolic versus exponential model, the percentage of variance accounted for (R2) between the fitted function and the actual subjective value of each delayed reward were compared for each model. Wilcoxon signed-ranks tests were conducted to assess the significance of the within-group differences in R2 for the hyperbolic and exponential models for current smokers, never-smokers, and ex-smokers.
Results Table 1 shows demographic and participant characteristics as well as Eysenck Impulsivity Scores for the current smokers, never-smokers, and ex-smokers. There were no significant differences in demographic or participant characteristics. The one-way ANOVA of the Eysenck Impulsivity Score revealed an overall effect of group that approached significance [F(2, 63)=2.55, P=0.09]. Subsequent planned pairwise comparisons determined that scores for smokers differed significantly from scores for ex-smokers (t=2.23, P=0.03), but scores for smokers versus never-smokers and scores for never-smokers versus ex-smokers did not differ significantly. Figure 1 shows the median indifference points for the monetary delay discounting assessment for current smokers, never-smokers, and ex-smokers at each of seven delays. The functions show the estimated value of the immediate reward at the point of subjective equality at a particular delay. The hyperbolic functions were fit to the median indifference points. The subjective value decreased more rapidly for smokers than for never-smokers and ex-smokers. For example, for smokers $1000 lost 42.5% of its value when delayed by 1 year, but for never- and ex-smokers, $1000 lost only 17.5% of its value when delayed by 1 year. These differences were assessed statistically by comparing the estimated discounting parameter (k) obtained by fitting the hyperbolic function to data from each individual in each group. The median k for money for smokers, never-smokers, and ex-smokers was 0.05414, 0.0073, and 0.0071, respectively. A Kruskal-Wallis one-way ANOVA revealed a significant
451 R2
Table 2 Median for the hyperbolic and exponential models of delay discounting for current, never, and ex-smokers for monetary outcomes and for current smokers for cigarette outcomes Outcome
Fig. 1 Temporal discounting functions for money for current smokers (squares) never-smokers (triangles) and ex-smokers (circles). Points show median indifference points for money as a function of delay. Lines show the best-fitting discounting functions generated by the hyperbolic decay model (see text)
Fig. 2 Temporal discounting functions for money (squares) and cigarettes (diamonds) for current smokers. Points show median indifference points as a function of delay. Lines show the best-fitting discount functions generated by the hyperbolic decay model (see text)
overall effect of group on the monetary discounting parameter (Kruskal-Wallis=11.84, P=0.003). Subsequent pairwise comparisons determined that the discounting parameter for smokers differed significantly both from the discounting parameter for never-smokers (U=133, P=0.007), and the discounting parameter for ex-smokers (U=107, P=0.002). The discounting parameter for neversmokers and ex-smokers did not differ significantly. Figure 2 shows the median indifference points at each of seven delays for current smokers for the cigarette delay discounting assessment and for the monetary discounting assessment (reproduced from Figure 1 for ease of comparison). The points represent the value of the immediate outcome (either cigarettes or money) when the current smokers were indifferent between the immediate and fixed-amount delayed outcomes. The lines show the hyperbolic functions fit to the median indifference points. The function decreases more steeply for cigarettes than for money. For example, $1000 worth of cigarettes lost 87.5% of its value when delayed by 5 years, but $1000 lost only 57.5% of its value when delayed by 5 years. These differences were assessed statistically by comparing the estimated discounting parameter (k) from
R2 Exponential
Hyperbolic
Money Current smokers Never-smokers Ex-smokers
0.74 0.81 0.83
0.89 0.88 0.85
Cigarettes Current smokers
0.87
0.89
each individual smoker for each outcome type. The median k for money was 0.0541 and for cigarettes was 0.0636. The Wilcoxon signed-ranks test confirmed that for current smokers, the discounting parameter for money differed significantly from the discounting parameter for cigarettes (P=0.049). Table 2 shows the median R2 for the hyperbolic and exponential functions for the monetary delay discounting assessment for all groups and for the cigarette delay discounting assessment for current smokers. For the monetary assessment, the hyperbolic function provided a better fit (higher R2) to the individual data for 21 out of 23 current smokers, 21 out of 22 never-smokers, and 17 out of 21 ex-smokers. Wilcoxon signed-ranks tests confirmed R2 for the exponential and hyperbolic models was significantly different for current smokers (P=0.0002), never-smokers (P=0.0007) and ex-smokers (P=0.003). For the cigarette delay discounting assessment, the hyperbolic function provided a better fit (higher R2) to the individual data for 15 out of 23 current smokers. The Wilcoxon signed-ranks test confirmed the difference in R2 was statistically significant (P=0.018).
Discussion There were three major findings in the present study. First, current smokers were found to discount monetary outcomes to a greater extent than did never or ex-smokers. Interestingly, never- and ex-smokers did not differ in their discounting of delayed outcomes. Second, for current smokers, cigarettes were discounted to a greater extent than monetary outcomes. Third, the hyperbolic equation (Equation 2) fit the data better than the exponential discounting equation (Equation 1). There are six points related to these findings that we wish to discuss. First, the finding that smokers discounted monetary outcomes to a greater extent than did control participants is consistent with those of prior studies showing that individuals with drug abuse problems discount monetary outcomes more steeply than do control participants (e.g., Madden et al. 1997; Vuchinich and Simpson 1998). An important aspect of extending this observation to smokers is that they often do not experience the life-disrupting
452
events (e.g., homelessness, poverty) evident in other drug-dependent populations. This result suggests a greater a degree of discounting may be more closely related to drug dependence disorders than with the attendant problems experienced by certain drug dependent groups (e.g., heroin-dependent individuals). Examining delay discounting in other populations who experience drug dependence but do not necessarily experience these additional problems (e.g., middle-class alcoholics) could help to evaluate this hypothesis. Second, an important element of this study was the degree of discounting by ex-smokers. Ex-smokers discounted monetary outcomes less sharply than did current smokers and to a similar extent as never-smokers. This result suggests that the increased degree of discounting found in this study for current smokers could be a reversible effect of nicotine dependence. According to this interpretation, chronic cigarette smoking could produce steeper discounting, and upon smoking cessation, the degree of discounting may decrease. However, the similarities in discounting between never-smokers and ex-smokers could also result from a selection bias: perhaps only smokers with a relatively low degree of discounting are able to successfully quit smoking cigarettes and thereby become ex-smokers. Only a longitudinal study which measured the degree of discounting in cigarette smokers who successfully quit smoking compared to those who did not, and which measured the degree of discounting both prior to and after abstinence for ex-smokers, could distinguish between the two interpretations offered above. However, the frequency distributions of k in the present study could offer a tentative interim exploration of whether the similar degree of discounting observed in the present study between never- and ex-smokers is due to reversible changes in discounting or to selection bias. Both interpretations predict the frequency distributions of k to be similar for never- and ex-smokers. If the similarity were due to reversible changes in discounting, the frequency distributions of k for current smokers could be shifted uniformly to the right (toward higher values) compared to never- and ex-smokers. However, if the similarity were due to selection bias, and only smokers with a relatively low degree of discounting are able to successfully quit, then the distribution for current smokers could be bimodal. One mode would be a cluster of relatively low discounting parameters, contributed by future ex-smokers and therefore similar to that for neverand ex-smokers, and the other mode would be a concentration of relatively high discounting parameters contributed by smokers unlikely to successfully quit. Figure 3 shows the distribution of the discounting parameter (k) for current smokers is most consistent with the selection bias interpretation of the similarities between the degree of discounting for never- and ex-smokers. The relative frequency distributions for never- and ex-smokers are similar, whereas the distribution for current smokers, while showing a similar cluster around low k values, also shows clustering at higher k values. The
Fig. 3 Relative frequency distributions of k (the delay discounting parameter) estimated by the hyperbolic model for current, never, and ex-smokers (see text)
lower cluster of discounting rates for current smokers could be contributed by smokers who are likely to successfully quit (i.e., become ex-smokers), and the higher clustering could be contributed by smokers who are unlikely to successfully quit. However, this interpretation of the similarities between the degree of discounting for never- and ex-smokers should be viewed with caution. The sample sizes in the current study were not large, and the analysis was conducted post-hoc and could be artifactual. Whether this analysis would be supported by a longitudinal study of these issues is not clear. Third, the observation that smokers discount cigarettes more than money systematically replicates prior studies in which heroin dependent individuals were shown to discount heroin more so than monetary outcomes (Madden et al. 1997; Odum et al., unpublished data). Collectively, these findings suggest that the degree of discounting varies by reward type and appears to reflect the importance of drug reward relative to money. One potential reason for this drug-specific steep discounting is that both nicotine and opioid dependence are marked by withdrawal. In particular, nicotine withdrawal symptoms include dysphoria, irritability, anxiety, difficulty concentrating, restlessness, insomnia, and increased appetite (Hughes et al. 1991). To avoid these symptoms, nicotine in some form must be consumed regularly. Consistent with this speculation, laboratory work suggests that humans may behave more impulsively when attempting to escape from aversive events (e.g., Solnick et al. 1980; Navarick 1982). Investigating the role of withdrawal processes in the discounting of delayed rewards may entail examining drug dependencies that do not have as pronounced withdrawal symptoms (e.g., cocaine dependence) and actively manipulating drug deprivation and hence withdrawal to examine its effect on discounting. Fourth, this study demonstrated that in most cases the hyperbolic discounting equation better fit the data than the exponential discounting equation. This finding replicates prior demonstrations of the superior fit of the hyperbolic function in describing discounting by non-drugusing participants (e.g., Rachlin et al. 1991; Rachlin and
453
Raineri 1992; Kirby and Marakovic 1995; Myerson and Green 1995; Kirby 1997) or college students with undiagnosed alcohol problems (Vuchinich and Simpson 1998). These studies examined how these equations fit data derived from monetary outcomes. In a prior study with opioid-dependent patients we demonstrated the hyperbolic equation better fit the data obtained for money and heroin outcomes (Madden et al. 1999). The present study extends the generality of hyperbolic discounting to another clinically diagnosed population of drug-dependent individuals and to choices involving another type of drug reward. Understanding that delayed rewards, particularly in drug dependent populations, are discounted with the extremely bowed functions typical of the hyperbolic equation permits us to understand why drug dependent individuals will engage in actions that put future rewards at risk and why efforts to convince drug dependent individuals to be concerned with their future are generally so ineffective. Indeed, the efficacy of contingency management programs (Higgins et al. 1994; Roll et al. 1996; Bickel et al. 1997) may be in large part due to the relatively immediate reinforcement for abstinence. If the reinforcement were delayed, contingency management could be quite ineffectual. An interesting study would be to determine if an individual’s degree of discounting determines the reinforcement delay that would be effective in reinforcing abstinence. Fifth, one potential criticism of the present study is that participants made choices about hypothetical monetary and cigarette outcomes, rather than actually receiving the outcomes. Although the results from these procedures must be viewed with caution, there several reasons why they could have merit. First, results from delay discounting experiments using hypothetical outcomes with humans, such as the present, yield similar hyperbolic functions to those using real outcomes, such as food and water, with non-humans (e.g., Mazur 1987; Richards et al. 1997). Furthermore, the hyperbolic equation has been shown to describe accurately the behavior of humans choosing real (Kirby and Marakovic 1995) and hypothetical (Rachlin et al. 1991) monetary outcomes as well as real (Kirby and Herrnstein 1995) and hypothetical consumer goods (Raineri and Rachlin 1993). Furthermore, the hypothetical money choice task has been used in over a dozen published experiments, with a variety of populations, all finding hyperbolic discounting. Spurious results seem unlikely to all conform to a hyperbolic function. Finally, hypothetical outcomes can allow the laboratory investigation of factors that would be difficult to study otherwise with humans for practical or ethical reasons (Petry and Bickel 1998). Sixth and finally, different degrees of discounting future outcomes could arise in a number of ways. Some evidence suggests steep discounting could occur prior to (and possibly precipitate) problems of drug abuse. In a prospective longitudinal study, adolescents with impulse control problems (as assessed by questionnaire) at an early age were more likely to have developed drug abuse problems when retested at a later age (Shedler and Block
1990). Alternatively, drug abuse could occur prior to (and possibly precipitate) little regard for relatively longterm events. As an additional possibility, another factor could produce both drug abuse and rapid discounting of future outcomes. To determine which of these possibilities is correct will obviously require more research. Whatever the direction of causality though, discounting may be malleable: procedures have been shown to affect impulsivity (i.e., choice of a smaller, more immediate reinforcer over a larger, more delayed reinforcer; Ainslie 1974) in children (e.g., Schweitzer and Sulzer-Azaroff 1988) as well as in non-humans (Mazur and Logue 1978). Acknowledgements This research was supported in part by grants T32DA07242, R37DA06526 and R01DA11692 from the National Institute on Drug Abuse (NIDA). The authors thank Timothy Shahan for helpful comments and Margaret Tillile for research assistance.
References Ainslie G (1974) Impulse control in pigeons. J Exp Anal Behav 21:485–489 Ainslie G (1992) Picoeconomics: the strategic interaction of successive motivational states within the person. Cambridge University Press, Cambridge Ainslie G, Haslam N (1992) Hyperbolic discounting. In: Loewenstein G, Elster J (eds) Choice over time. Russell Sage Foundation, New York, pp 57–92 Alcock CC, Grace DM (1988) Pathological gamblers are neither impulsive nor sensation-seekers. Aust NZ J Psychiatry 22: 307–311 Becker GS, Murphy KM (1988) A theory of rational addiction. J Pol Econ 96:675–700 Bickel WK, Amass L, Higgins ST, Badger GJ, Esch RA (1997) Effects of adding behavioral treatment to opioid detoxification with buprenorphine. J Consult Clin Psychol 65:803–810 Eysenck SBB, Eysenck HJ (1978) Impulsiveness and venturesomeness: their position in a dimensional system of personality description. Psychol Rep 43:1247–1255 Fishburn PC, Rubinstein A (1982) Time preference. Int Econ Rev 23:677–694 Geist CR, Herrmann SM (1990) A comparison of the psychological characteristics of smokers, ex-smokers, and nonsmokers. Clin Psychol 46:102–105 Green L, Fry AF, Myerson J (1994) Discounting of delayed rewards: a lifespan comparison. Psychol Sci 5:33–36 Higgins ST, Budney AJ, Bickel WK, Foerg FE, Donham R, Badger GJ (1994) Incentives improve outcome in outpatient behavioral treatment of cocaine dependence. Arch Gen Psychiatry 51:568–576 Hughes JR (1993) Smoking is a drug dependence: a reply to Robinson and Pritchard. Psychopharmacology 113:282–283 Hughes JR, Gust SW, Skoog K, Keenan RM, Fenwick JW (1991) Symptoms of tobacco withdrawal. Arch Gen Psychiatry 48: 52–59 Kirby KN (1997) Bidding on the future: evidence against normative discounting of delayed rewards. J Exp Psychol Gen 126: 54–70 Kirby KN, Herrnstein RJ (1995) Preference reversals due to myopic discounting of delayed reward. Psychol Sci 6:83–89 Kirby KN, Marakovic NN (1995) Modeling myopic decisions: evidence for hyperbolic delay-discounting within subjects and amounts. Organ Behav Hum Decis Proc 64:22–30 Kirby KN, Petry NM, Bickel WK (1999) Heroin addicts have higher discount rates for delayed rewards than non-drug using controls. J Exp Psychol Gen 128:78–87
454 Lancaster K (1963) An axiomatic theory of consumer time preference. Int Econ Rev 4:221–231 Lipkus IM, Barefoot JC, Williams RB, Siegler IC (1994) Personality measures as predictors of smoking initiation and cessation in the UNC Alumni Heart Study. Health Psychol 13: 149–155 Logue AW (1988) Research on self-control: an integrating framework. Behav Brain Sci 11:665–709 Logue AW (1995) Self-control. Prentice Hall, Englewood Cliffs, N.J. Madden GJ, Petry NM, Badger GJ, Bickel WK (1997) Impulsive and self-control choices in opioid-dependent patients and nondrug-using control participants: drug and monetary rewards. Exp Clin Psychopharmacol 5:256–262 Madden GJ, Bickel WK, Jacobs EA (1999) Discounting of delayed rewards in opioid-dependent outpatients: exponential or hyperbolic discounting functions? Exp Clin Psychopharmacol (in press) Mazur JE (1987) An adjusting procedure for studying delayed reinforcement. In: Commons ML, Mazur JE, Nevin JA, Rachlin H (eds) Quantitative analysis of behavior: vol 5. The effect of delay and intervening events on reinforcement value. Erlbaum, Hillsdale, N.J., pp 55–73 Mazur JE, Logue AW (1978) Choice in a “self-control” paradigm: effects of a fading procedure. J Exp Anal Behav 30: 11–17 Meyer RF (1976) Preferences over time. In: Keeney RL, Raiffa H (eds) Decisions with multiple objectives: preferences and value tradeoffs. Wiley, New York, pp 473–514 Mitchell SH (1999) Personality and behavioral measures show that cigarette smokers are more impulsive than never-smokers [Abstract]. In: Harris LS (ed) Proceedings of the College on Problems of Drug Dependence 1998, NIH Publication Number 99-4395, NIDA Research Monograph 179. Superintendent of Documents, US Government Printing Office, Washington, D.C., p 191 Myerson J, Green L (1995) Discounting of delayed rewards: models of individual choice. J Exp Anal Behav 64:263–276 Navarick DJ (1982) Negative reinforcement and choice in humans. Learn Motiv 13:361–377
Petry NM, Bickel WK (1998) Polydrug abuse in heroin addicts: a behavioral economic analysis. Addiction 93:321–335 Rachlin H, Raineri A (1992) Irrationality, impulsiveness, and selfishness as discount reversal effects. In: Loewenstein G, Elster J (eds) Choice over time. Russell Sage Foundation, New York, pp 93–118 Rachlin H, Raineri A, Cross D (1991) Subjective probability and delay. J Exp Anal Behav 55:233–244 Raineri A, Rachlin H (1993) The effect of temporal constraints on the value of money and other commodities. J Behav Decis Making 6:77–94 Richards JB, Mitchell SH, de Wit H, Seiden LS (1997) Determination of discount functions in rats with an adjusting-amount procedure. J Exp Anal Behav 67:353–366 Rodriguez ML, Logue AW (1988) Adjusting delay to reinforcement: comparing choice in pigeons and humans. J Exp Psychol [Anim Behav Proc] 14:105–117 Roll JM, Higgins ST, Badger GJ (1996) An experimental comparison of three different schedules of reinforcement of drug abstinence using cigarette smoking as an exemplar. J Appl Behav Anal 29:495–504 Rosenthal TL, Edwards NB, Ackerman BJ, Knott DH, Rosenthal RH (1990) Substance abuse patterns reveal contrasting personality traits. J Subst Abuse 2:255–263 Schweitzer JB Sulzer-Azaroff B (1988) Self-control: teaching tolerance for delay in impulsive children. J Exp Anal Behav 50:173–185 Shedler J, Block J (1990) Adolescent drug use and psychological health: a longitudinal inquiry. Am Psychol 45:612–630 Solnick JV, Kannenberg CH, Eckerman DA, Waller MB (1980) An experimental analysis of impulsivity and impulse control in humans. Learn Motiv 11:61–77 Stevens SS (1975) Psychophysics. Wiley, New York Vuchinich RE, Simpson CA (1998) Hyperbolic temporal discounting in social drinkers and problem drinkers. Exp Clin Psychopharmacol 6:292–305 Waldeck TL, Miller LS (1997) Gender and impulsivity differences in licit substance use. Subst Abuse 9:269–275 Webster CD, Jackson MA (1997) Impulsivity: theory, assessment and treatment. Guilford Press, New York
