Cigarette Smokers Discount Past and Future Rewards Symmetrically and More than Controls: Is Discounting a Measure of Impulsivity?

1. Introduction

Considerable evidence supports the overlap of processes and brain regions associated with consideration of the past and anticipation of the future (see Buckner and Carroll, 2007, for a review). Recent observations of Hassabis and colleagues demonstrated that amnesic patients with bilateral damage to the hippocampus were impaired in their ability to imagine new personal experiences in addition to being unable to remember their past (Hassabis et al., 2007). Numerous other studies (e.g., Addis et al., 2007; Atance and O'Neill, 2001; Atance and O'Neill, 2005; Dretsch and Tipples, 2008; Okuda et al., 2003) of diverse populations and procedures support the hypothesis that behaviors associated with the past and future are closely related.

One research area that has extensively and exclusively focused on the future is temporal discounting (see Frederick et al., 2002; Green and Myerson, 2004, for reviews). Temporal discounting, often considered a measure of impulsivity, refers to the decrease in the subjective value of an outcome as delay to its availability increases (Green and Myerson, 2004). Studies of temporal discounting indicate that future outcomes are discounted hyperbolically, that the degree of discounting decreases as the magnitude of the reward increases in humans (magnitude effect), and that losses tend to be discounted less than gains (sign effect).

Impulsivity is inferred from the perspective of temporal discounting when a smaller, more immediate reward is preferred to a later, larger one. Among the populations that exhibit greater discounting of future rewards relative to matched control include: children with attention deficit and hyperactivity disorder (Schweitzer and Sulzer-Azaroff, 1995), cigarette smokers (Bickel et al., 1999), heroin addicts (Madden et al., 1997), cocaine-dependent individuals (Heil et al., 2006), and pathological gamblers (see Bickel and Marsch, 2001; Dixon et al., 2003; Reynolds, 2006, for reviews). The greater discounting (preference for the sooner smaller reward) exhibited by these populations has been considered prima facie evidence of impulsivity; that is, gratification was not deferred and/or preference for the immediate reward was not inhibited.

Recently, our group has extended this paradigm to the discounting of past outcomes (Yi et al., 2006). The results from that initial study demonstrated that past outcomes were discounted hyperbolically by undergraduate college students (e.g., they valued rewards in the near past more than rewards further back in the past), that small rewards were discounted more than large rewards (the magnitude effect), and rewards were discounted more than losses (sign effect). Moreover, the degree of past discounting was significantly correlated with future discounting. These results were not intuitive. From the standpoint of traditional economics, a preference might be expected for the largest amount in the distant past that would permit greater investment returns in the present. Another viewpoint is that the past is over, has led to the current condition, and therefore systematic valuing of the past is nonsensical. Instead, we found discounting of past and future outcomes were closely related, and past discounting tended to follow the functional relations observed in future discounting.

The present study is the first to compare temporal discounting of past and future outcomes among an addicted population, cigarette smokers, relative to controls. We will also examine whether the effects of magnitude and sign (gains, losses) observed in future discounting operate symmetrically to past discounting. Potentially, there are two outcomes: symmetry and asymmetry of past and future. First, if cigarette smokers discount future and past rewards more than controls, this finding will support the growing recognition of the close and possibly symmetric relationship between past and future. Moreover, this observation would also challenge the view that discounting is a measure of impulsivity since discounting of events in the past does not require the inhibition of responding or the delay of gratification. Second, we could find a lack of future and past discounting symmetry comparing smokers and non-smokers. If this result were observed, it would undermine or provide a limiting case about the relationship between the past and the future. A lack of symmetry would support the notion that future discounting is linked to impulsivity and that past discounting may be germane to other phenomena.

2. Methods

3. Results

4. Discussion

In this study, we examined the discounting of future and past gains and losses by cigarette smokers and controls. Overall, we found that the pattern of results for future and past outcomes was generally symmetrical for both smokers and controls, and cigarette smokers discounted both future and past hypothetical monetary gains significantly more than controls. These data demonstrate that the temporal construal of gains extends both into the past and the future. The lack of difference in discounting of past and future losses between smokers and controls, and the weaker correlations between past and future losses (relative to gains) may indicate potential differences between discounting gains and losses. Four points regarding the obtained results follow.

First, hyperbolic discounting, the magnitude effect, and population differences (smokers vs. controls) supports symmetry between future and past discounting of monetary gains. As in many previous studies of future discounting (e.g., Kirby, 1997) and in the one prior study of past discounting (Yi et al., 2006), the hyperbolic model of discounting proposed by Mazur (1987) proved a superior model of human discounting behavior. Furthermore, small-magnitude gains were discounted more than large-magnitude gains (the magnitude effect). This has been well demonstrated with money (hypothetical and real) and with hypothetical drug reinforcers (Baker et al., 2003; Giordano et al., 2002) among humans (drug dependent and human controls). The extension of the magnitude effect to the past and to cigarette smokers supports the notion that that processes and brain regions that value different magnitudes of reward may act similarly when considering both past and future rewards, affected comparably by the conditions associated with dependence on tobacco. The sign effect, where gains are discounted more than losses, was not replicated in the present experiment in either group, though non-significant differences were in the predicted direction. Future studies using these procedures in conjunction with neuroimaging techniques would be able to clarify the co-identity of brain regions associated with past and future discounting.

Second, greater preference for smaller, more immediate rewards over larger, more delayed rewards has been considered an indication of impulsivity; that is, the individual was unwilling or unable to defer gratification or inhibit responding. Although impulsivity can apply to a wide range of measures and observations (Evenden, 1999), the fact that cigarette smokers discount past gains more than controls presents some challenges to the notion that discounting refers to inability to inhibit responding or delay gratification; there is nothing to inhibit in the past. Indeed, to insure that all choices were referential to the past is precisely why we selected an hour ago as the more immediate option in the past discounting condition of our study.

What is likely is that that the same valuation and discounting processes are engaged when considering events in the future or in the past, and those who are dependent on substances are prone to be more focused on a temporally limited “now.” Consider that the experience of “now” likely includes events that happened in the immediate past and are anticipated to occur in the immediate future. Thus, the discounting of future and past rewards, rather than measuring the inability to inhibit responding, may define a temporal window that extends symmetrically from the present into both the future and the past. This smaller temporal window of high-discounting individuals suggests that cigarettes smokers may be more “stuck in time” (Roberts, 2002). The implications of this will have to be further examined, but may indicate a mechanism that explains why addicts seem to neither learn from prior experience nor anticipate the future. Moreover, these findings are also consistent with competing neurobehavioral systems hypotheses of addiction (Bechara, 2005; Bickel et al., 2007). These hypotheses suggest that addiction results from hyperactive impulsive systems exhibited by mid-brain systems related to reinforcement (e.g., limbic region) and hypoactive executive systems (e.g., prefrontal cortex). Given the principal function of the executive system and prefrontal cortex to cross-temporally organizanize behavior (Fuster, 1985; Fuster, 1997; Miller and Cohen, 2001), a diminished or hypoactive executive system and prefrontal cortex could result in the restricted temporal focus frequently observed among the addicted.

Third, the present study replicates the prior findings of greater discounting of future gains among cigarettes smokers (Bickel et al., 1999) and extends those results to discounting of past gains. We replicate a substantial number of studies that have demonstrated that adult cigarette smokers discount future gains more than controls (e.g., Baker et al., 2003; Heyman and Gibb, 2006; Mitchell, 1999; Ohmura et al., 2005; Reynolds et al., 2004). We also provide the first demonstration that nicotine-dependent individuals discount past gains more than controls. Consistent with the first published study of past discounting (Yi et al., 2006), this finding suggests that valuation is not only operative in considering future actions, but applies to the valuation of past events. What remains to be empirically determined, is whether other forms of dependence also demonstrate greater discounting of past outcomes, and if events that have been documented to alter discounting of the future (such as withdrawal; Giordano et al., 2002) also produce symmetrical changes in past discounting.

Fourth, observed results in the discounting of losses were somewhat equivocal. Though smokers discounted future and past losses more than nonsmokers, this difference did not reach statistical significance. This is partially consistent with Baker et al. (2003), who found that cigarette smokers discount losses more than nonsmokers, and Ohmura et al. (2005), who found no differences in the discounting of losses. Given the relatively limited between-condition variability observed in the discounting of losses (Estle et al., 2006), this is not surprising. Importantly, the same profile of results was observed in both future and past discounting of losses, maintaining future / past symmetry.

Finally, this study was conducted using only hypothetical events, which may have influenced the findings. A growing body of literature has compared discounting of real and hypothetical future rewards as the focus of the study or as one condition of a larger study (Johnson and Bickel, 2002; Lagorio and Madden, 2005; Madden et al., 2003). These studies all demonstrate very similar findings when comparing identical magnitudes. Comparisons of real and hypothetical outcomes in the past, unlike the future, are procedurally challenging, if not impossible. As a consequence, direct comparison of real and hypothetical may not be possible. Fortunately, discounting is a phenomenon that has functional relations (e.g., hyperbolic in form, magnitude effect) that are well established and replicated. Thus, if the results of the hypothetical were spurious, it is difficult to imagine why the results would follow the contours of those functional relationships. Moreover, arguably there are multiple ways to interpret the past discounting procedures. The greater variability in past losses may indicate between-subject variability in the interpretation of the choice. Therefore, there may be value of debriefing participants and asking how the participants interpreted and framed the past-discounting choices they were presented.

In conclusion, the present study extends the use of novel discounting procedure to cigarette smokers. Cigarette smokers discount the future and past more than controls, providing compelling support of recent empirical findings and theoretical frameworks suggesting that the future and past may be causally linked. The ability to consider and value the future may be tied to the ability to value the past. With respect to addiction, this research opens up the possibility of new lines of inquiry examining whether variables that affect future discounting also affect past discounting. Similarly, there may be value in considering whether phenomena tied to the past such as memory could be used to change future discounting.