Is discounting impulsive?: Evidence from temporal and probability discounting in gambling and non-gambling college students
Introduction
The present study focuses on the relationship between impulsivity, delay of gratification, and risk taking, and how these constructs relate to performance on temporal and probability discounting tasks. Does behavior on discounting tasks measure an underlying general trait of impulsivity? That is, are those people whose preferences are especially affected by the amount of time until receipt of rewards also the ones whose preferences are relatively less affected by the probability of not receiving a reward? Alternatively, do temporal and probability discounting tasks assess relatively independent behavioral characteristics and reflect separate traits (Green et al., 1999, Mitchell, 1999, Myerson et al., in press)?
One way to address this issue is to examine correlations between behavioral measures of discounting. For example, if performance on temporal and probability discounting tasks both reflect the same general trait of impulsivity, then one would expect that those who discount delayed rewards more steeply also would show shallower discounting of probabilistic rewards. That is, impulsive individuals might be thought of as showing steep temporal discounting, in that they are less influenced by the long-term consequences of their behavior, and also as showing shallow probability discounting, in that they are less affected by uncertainty. Thus, if impulsivity underlies discounting of both delayed and probabilistic rewards, then a negative correlation between measures of temporal and probability discounting should be observed.
Following this line of reasoning, Myerson et al. (in press) recently examined the degree to which individuals discounted small and large delayed and probabilistic rewards. Myerson et al. failed to find evidence of the negative correlation expected if impulsivity were a general trait underlying performance on both temporal and probability discounting tasks. Indeed, in two large samples the correlations between the degree of temporal and probability discounting were consistently positive (although not always significant). Positive correlations between temporal and probability discounting measures also have been reported by Richards et al. (1999).
Another approach to studying the relationship between impulsivity and risk taking might be to look at extreme groups. For example, one could study groups who engage in risky behavior and see whether their behavior is also impulsive in other ways. One might think of recent research on discounting by substance abusers as bearing on this issue. Substance abusers tend to discount delayed rewards more steeply than controls (e.g. Kirby et al., 1999, Madden et al., 1997, Petry, 2001a), but it is unclear whether substance abusers also discount probabilistic rewards less steeply than controls. The only study to examine probability discounting in substance abusers found no group differences in probability discounting, although smokers did discount delayed rewards more steeply than controls (Mitchell, 1999). Another potentially interesting use of an extreme group’s approach to these issues would be to study gamblers and non-gamblers. Gamblers may be thought of as being impulsive because they engage in behavior despite the possible negative long-term consequences. Moreover, gambling, by definition, involves risk because the outcome is uncertain.
In addition to its theoretical relevance, gambling is also of interest because of its social implications. Gambling represents both a major social problem as well as a personal problem for many individuals. According to a recent report to the National Gambling Impact Study Commission (Gerstein et al., 1999), there are an estimated 5.5 million pathological and problem gamblers in the US, with economic costs estimated to be at least $5 billion annually. Moreover, gambling in the US appears to be increasing. According to a recent survey, more than 80% of adults now engage in some form of gambling annually (Welte et al., 2002). Many of these people may be at risk for developing serious gambling problems that frequently manifest themselves only after years of more controlled gambling (American Psychiatric Association, 1994).
Consistent with the view that gambling reflects impulsivity, and that impulsivity is a general trait that involves not only risk taking but also a difficulty in delaying gratification, Petry (2001b) found that pathological gamblers discounted delayed rewards more steeply than controls, although she did not examine the discounting of probabilistic rewards. It might be argued, however, that pathological gamblers (as defined by the DSM-IV, the Diagnostic and Statistical Manual, 4th ed.; American Psychiatric Association, 1994) are a special case, and by definition, are not representative of gamblers as a whole. For example, pathological gamblers show high rates of comorbidity with other psychiatric disorders (e.g. Blaszczynski and Steel, 1998, Ibanez et al., 2001) as well as a host of health and other problems (Gerstein et al., 1999). In the present context, such co-occurring problems may complicate interpretation of data from pathological gamblers.
In addition to comorbidity issues, pathological gamblers are problematic to study for other reasons. They not only differ from non-gamblers with respect to recent gambling behavior, but they also are likely to have a history of interpersonal and financial problems resulting from their gambling. Attempts to match non-gambling controls on such characteristics may require introducing new differences. For example, in order to control for the financial pressures facing pathological gamblers, Petry (2001b) selected non-gamblers whose median annual income was less than half that of the gamblers.
In order to avoid some of these problems, particularly those associated with a history of interpersonal and financial problems, the present study examined the performance of gambling and non-gambling college students on temporal and probability discounting tasks. In an earlier pilot study that examined discounting in convenience samples of college students, we (Holt et al., 2001) compared rates of temporal and probability discounting from students who reported that they gambled three or more times a month with students who reported never gambling. Gamblers appeared to discount probabilistic rewards to a lesser degree than the non-gamblers, but gamblers and non-gamblers did not differ significantly in terms of their discounting of delayed rewards. This suggests a possible dissociation of the tendency to discount delayed rewards and the tendency to discount probabilistic rewards that, as such, would be inconsistent with the notion of a general impulsivity trait underlying both temporal and probability discounting.
The present study followed up on these findings using a more rigorous definition of who qualified as a gambler. For current purposes, gamblers were defined as those scoring 4 or higher on the South Oaks Gambling Screen (SOGS; Lesieur and Bloom, 1987), a self-report measure for assessing gambling frequency and related problems. Their performance on temporal and probability discounting tasks was compared to non-gambling college students, defined as those with SOGS scores of 0 or 1.
Previous research with these discounting tasks has shown that the devaluation of future and uncertain rewards is well described by a hyperbola-like discounting function:where V is the subjective value of a delayed or probabilistic reward, A is the reward amount, X is the delay to or the odds against receipt of the reward, b is a parameter that (with s held constant) reflects the rate of discounting, and s is a non-linear scaling parameter (e.g. Green et al., 1999). When s=1.0, Eq. (1) reduces to a simple hyperbola (Mazur, 1987).
The present study is concerned with three issues. The first issue concerns how well a hyperbola-like discounting function, Eq. (1), describes discounting of both smaller and larger delayed and probabilistic rewards by both gamblers and non-gamblers. The second issue concerns whether there are group differences in temporal and probability discounting. The final issue concerns whether or not the tendency to discount delayed rewards is negatively correlated with the tendency to discount probabilistic rewards.
If a general trait of impulsivity were to underlie the discounting of both delayed and probabilistic rewards, then one might expect that individuals who have difficulty delaying gratification would also be more likely to gamble and take risks. Thus, if impulsivity underlies both temporal and probability discounting, then one would expect to see gamblers discount delayed rewards more steeply but discount probabilistic rewards less steeply than non-gamblers. The general trait hypothesis also would predict negative correlations between the degree of delay and probability discounting within each group as well as in the combined sample. The present study assessed the validity of these predictions.
Section snippets
Participants
Participants were 18- to 24-year-old college undergraduates at a state university in the Midwest. After providing informed consent for participation in a study of decision making, volunteers completed a demographic information sheet, the SOGS (Lesieur and Bloom, 1987), and a revised version of the Eysenck Personality Questionnaire (EPQ; Eysenck et al., 1985). The SOGS is a valid and reliable measure of self-reported gambling behavior for differentiating groups, although caution is called for in
Gambling and personality measures
The mean SOGS score for the gamblers was 6.53 (S.D.=3.12; median=5) whereas the mean score for the non-gamblers was 0.26 (S.D.=0.45; median=0). In addition to the items used in calculating an individual’s score, there are also several items on the SOGS that are not used in calculating the score. Importantly, one of these items asks the frequency with which an individual engages in each of 10 different gambling activities (e.g. played cards for money, went to casino). Thirteen of the 19
Discussion
The present findings reveal fundamental similarities between the discounting of delayed and probabilistic rewards by groups of gambling and non-gambling college students. A hyperbola-like function, Eq. (1), provided good descriptions of discounting by individuals in both groups, and amount of reward had opposite effects on the degree of discounting delayed and probabilistic rewards by both gamblers and non-gamblers. That is, both groups discounted large delayed amounts less steeply than small
Acknowledgements
Support for the current project was provided by Grant MH 55308 from the National Institutes of Health. We thank Professor Anthony Marcattilio for his assistance in coordinating the data collection.
References (35)
- et al.
A revised version of the Psychoticism scale
Pers. Individ. Diff.
(1985) - et al.
Excessive discounting of delayed rewards in substance abusers with gambling problems
Drug Alcohol Depend.
(1999) Reliability validity and classification accuracy of the South Oaks Gambling Screen (SOGS)
Addict. Behav.
(2002)- et al.
Pathological gamblers are neither impulsive nor sensation seekers
Aust. N.Z. J. Psychiatry
(1988) - American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders, 4th ed. Author,...
- Barratt, E.S., Stanford, M.S., 1995. Impulsiveness. In: Costello, C.G. (Ed.), Personality Characteristics of the...
- et al.
Personality disorders among pathological gamblers
J. Gambling Stud.
(1998) - et al.
Impulsivity in pathological gambling: the antisocial impulsivist
Addiction
(1997) - et al.
Cross-cultural comparisons of discounting delayed and probabilistic rewards
Psychol. Rec.
(2002) - et al.
Impulsiveness and venturesomeness: their position in a dimensional system of personality description
Psychol. Rep.
(1978)
The matching law and amount-dependent exponential discounting as accounts of self control choice
J. Exp. Anal. Behav.
Exponential versus hyperbolic discounting of delayed outcomes: risk and waiting time
Am. Zool.
Amount of reward has opposite effects on the discounting of delayed and probabilistic outcomes
J. Exp. Psychol. Learn. Mem. Cogn.
Psychiatric comorbidity in pathological gamblers seeking treatment
Am. J. Psychiatry
Heroin addicts have higher discount rates for delayed rewards than non-drug-using controls
J. Exp. Psychol. Gen.
Cited by (213)
The effects of early life stress on impulsivity
2022, Neuroscience and Biobehavioral ReviewsThe effect of delay and social distance in the dictator and ultimatum games
2021, Behavioural ProcessesEffects of Amount on Probability Discounting: A Replication and Extension
2021, Behavioural ProcessesCitation Excerpt :These individuals show relatively low rates of discounting, and they are also more likely to exhibit problem behavior. For example, low rates of probability discounting are associated with gambling (Holt et al., 2003), smoking (Yi et al., 2007), and risky sexual behavior (Johnson et al., 2016). However, low rates of discounting are not always indicative of problematic behavior.
Effects of economic context and reward amount on delay and probability discounting
2023, Journal of the Experimental Analysis of Behavior“Ten euros now” temporal discounting in Huntington disease
2023, Neurological SciencesReinforcer pathology, probabilistic choice, and medication adherence in patients with multiple sclerosis
2023, Journal of the Experimental Analysis of Behavior