Decision biases and persistent illicit drug use: an experimental study of distributed choice and addiction

https://doi.org/10.1016/S0376-8716(02)00071-6Get rights and content

Abstract

This experiment tested the hypothesis that differences in drug use are correlated with differences in decision making. The subjects were 22 drug clinic patients who had used either opiates or stimulants for an average of 10 years, and 21 community residents who reported that they had rarely used illicit addictive drugs. The procedure consisted of a series of binary choices with two consequences; they earned money and determined the intervals that separated choice trials. Each choice earned the same amount of money, but one initiated a shorter delay to the next trial, whereas the other initiated a shorter delay as averaged over the next two trials. Shorter delays were advantageous in that they increased the overall rate of earnings and they reduced the time spent waiting for the next trial. Thus, one choice was better from the perspective of the current trial, while the other choice was better from the perspective of two or more consecutive trials. Drug-clinic patients were more likely to favor the one-trial solution compared with control subjects, who were more likely to favor the two-trial solution. There were five different choice games, with different versions varying in the magnitude of the advantage for switching from the two-trial to the one-trial solution. Drug clinic and control subjects differed most in the games in which the immediate advantage of the one-trial solution was larger, and all subjects were more likely to choose the global solution when the incentive for switching to the one-trial solution was lower. The results support the view that individual differences in decision making influence the course of illicit drug use.

Introduction

In a paper on psychological and economic approaches to the study of choice, Herrnstein and Prelec (1991) emphasized a distinction that is especially pertinent to the study of addiction. They point out that some choices are distinct and unitary, whereas other choices are aggregates of ‘many smaller decisions, distributed over a period of time’. For instance, at a particular time one can decide to buy an exercise machine, whereas to be physically fit one must decide many times, over an extended period, to go exercise. This second case Herrnstein and Prelec aptly labeled ‘distributed choice’. Its relevance for the study of drug use is that addiction can be seen as an instance of distributed choice. One Friday night binge does not turn a social drinker into an alcoholic, and, similarly, a weekend without a drink does not turn an alcoholic into a teetotaler. Rather, addiction and recovery are states that reflect the cumulative effects of many small decisions.

The procedure for the experiment described in this report is based on distributed choice experiments that evaluated the predictive accuracy of economic and psychological models of behavior (‘maximizing versus matching law studies’, Herrnstein et al., 1993, Heyman, 1982, Vaughan, 1981). The subjects had two options on each trial. One gave a higher rate of return on the current trial; the other gave a higher rate of return overall. Or, put another way, there were two optimums: one, from the perspective of the current trial (a local optimum), and the other, from the perspective of two or more consecutive trials (the global optimum).

Elsewhere, it has been argued that the structure of this experiment is similar to the dilemma faced by drug users who are trying to abstain (Herrnstein and Prelec, 1992, Heyman, 1996). For example, it seems plausible that there are individuals for whom the reward value of a shot of heroin changes markedly as a function of the frame of reference. If the user considers the costs and benefits of competing activities in terms of a limited context, say the next few hours, heroin provides more value than does any other option. However, when the same user considers the advantages of different activities relative to a lifestyle, say a secure job and family versus the risks inherent in using an illegal substance on a daily basis, then its just the reverse, conventional activities trump drug use. Similarly, a smoker may prefer to have another cigarette if the frame of reference is the next few moments, but given the health risks of smoking, the same smoker may prefer a lifetime without cigarettes. Put more generally, under the conditions of distributed choice, preferences can systematically reverse as a function of whether the frame of reference is local or global.

Research on distributed choice problems with two or more optimums has been conducted with humans and non-humans. In experiments with pigeons, the typical non-human subject, the distribution of choices usually settled on the local optimum (e.g. Heyman and Herrnstein, 1986, Vaughan, 1981). However, pigeons could be taught to choose the global solution if the experiment included a stimulus that was correlated with higher overall reinforcement rates (Heyman and Tanz, 1995). In the initial experiments with humans, there was considerable individual variability, with some subjects stabilizing at the global optimum and others stabilizing at the local optimum (Herrnstein et al., 1993). Subsequently, Rachlin and his colleagues found that the temporal pattern of intertrial intervals influenced distributed choice. For example, if trials were presented three in a row followed by a pause, more subjects chose the global solution (Kudadjie-Gyamfi and Rachlin, 1996, Rachlin and Siegel, 1994). The pattern may have made trial-to-trial interactions (the global solution) more salient, which, in turn, suggests that individuals may differ in their sensitivity to the relationships between present and future consequences.

The present experiment investigated whether individual variation in distributed choice procedures was correlated with individual variation in drug use. In particular, we tested the hypothesis that drug clinic patients were more likely than control subjects to choose the local optimum in a series of distributed choice problems.

Section snippets

Subjects

Subjects were recruited from the North Charles Center for Addictions, McLean Hospital's Alcohol and Drug Abuse Treatment Unit, and neighborhoods near the two clinics. The North Charles clinic provides counseling services and methadone for opiate (usually heroin) addicts. The McLean drug treatment clinic is primarily a non-residential day program that provides counseling and daily drug screening. The methadone patients were tested at their counseling center, approximately 5 h after their daily

Choice results: global and local preferences

Fig. 1 summarizes performance in the five distributed choice games. The game numbers refer to the order in which each procedure occurred. On the x-axes, ‘3rds’ refers to the first, middle, and last third of a game. On the y-axes is the percentage of global responses for each third of the session. The average number of trials in each game was approximately 42. For panels one to five, the statistical analysis was conducted on the change in global response proportions in the middle and final third

Discussion

This study tested the hypothesis that individual differences in a distributed choice procedure would predict individual differences in drug use history. In accordance with the hypothesis, individuals with a history of long-term illicit drug use were more likely to choose the local solution than were subjects that did not have histories of long-term illicit drug use. ANCOVA revealed that the statistical significance of these differences remained after the groups were analytically equated for

Acknowledgements

This research was supported by a grant from the National Institute on Drug Abuse (R21 DA11954). We would like to thank Chris LaVancher for his expert help in developing the software for the distributed choice game and data collection.

References (21)

  • E. Kudadjie-Gyamfi et al.

    Temporal patterning in choice among delayed outcomes

    Organization. Behav. Hum. Decision Proc.

    (1996)
  • H. Rachlin et al.

    Temporal patterning in probabilistic choice

    Organization. Behav. Hum. Decision Proc.

    (1994)
  • D.A. Wasserman et al.

    Factors associated with lapses to heroin use during methadone maintenance

    Drug Alcohol Depend.

    (1998)
  • J.C. Anthony et al.

    Syndromes of drug abuse and dependence

  • R.J. Herrnstein et al.

    Utility maximization and melioration: internalities in individual choice

    J. Behav. Decision Making

    (1993)
  • R.J. Herrnstein et al.

    Melioration: a theory of distributed choice

    J. Economic Perspect.

    (1991)
  • R.J. Herrnstein et al.

    A theory of addiction

  • G.M. Heyman

    Is time allocation elicited behavior

  • G.M. Heyman

    Resolving the contradictions of addiction

    Behav. Brain Sci.

    (1996)
  • G.M. Heyman et al.

    More on concurrent interval-ratio schedules: a replication and review

    J. Exp. Anal. Behav.

    (1986)
There are more references available in the full text version of this article.

Cited by (0)

View full text