An Addiction Science Network Resource

Reprinted from H. de Wit and C.E. Johanson (1987), A drug preference procedure for use with human volunteers. In M.A. Bozarth (Ed.), Methods of assessing the reinforcing properties of abused drugs (pp. 559-572). New York: Springer-Verlag.
Brain Reward System
Back to book TOC
(Search book contents) (Search entire ASNet)


Chapter 26

A Drug Preference Procedure
for Use with Human Volunteers


H. de Wit and C. E. Johanson

The University of Chicago
Department of Psychiatry
Drug Abuse Research Center
5841 S. Maryland Avenue
Chicago, Illinois 60637

A choice procedure for measuring the reinforcing properties of drugs in humans is described. The use of humans as research subjects is contrasted to the use of laboratory animals in research with psychoactive drugs, and choice measures are contrasted to other measures of reinforcing efficacy. The procedure used in this laboratory is described in detail, and results of several recent studies are summarized. Methodological issues are also discussed.



Laboratory studies of behavior maintained by drugs as reinforcers in animals have gained acceptance as valid experimental models for studying human drug abuse (Thompson & Unna, 1977). This acceptance is based on studies showing that drugs that are abused by humans are also effective reinforcers of behavior in laboratory tests with animals (for reviews see Brady, Griffiths, Hienz, Ator, Lukas, & Lamb, this volume; Johanson & Balster, 1978; Johanson & Schuster, 1981; Weeks & Collins, this volume; Yanagita, this volume). Laboratory animals readily acquire and maintain operant responses when these are followed by delivery of drugs that have high dependence potential in humans (e.g., amphetamine or heroin). In contrast, drugs that are not abused, such as chlorpromazine, do not serve as effective reinforcers in laboratory animals. The study of pharmacological, behavioral, and organismic factors affecting the reinforcing properties of drugs in the laboratory will improve our understanding of factors influencing human drug-taking behavior.

Humans as Subjects

Whereas most studies on the reinforcing properties of drugs have utilized laboratory animals such as rats and monkeys as subjects (Griffiths, Bigelow, & Henningfield, 1980a; Johanson & Schuster, 1981), several researchers have used similar procedures with human subjects. Since work in other areas of behavioral research has shown inter-species generality, it is not surprising that the studies with humans have produced results similar to those with laboratory animals. For example, humans will perform an operant response at a high rate to obtain deliveries of drugs such as pentobarbital (Griffiths, Bigelow, & Liebson, 1976) or alcohol (Mello & Mendelson, 1971, this volume). The patterns of responding observed when a drug is delivered under schedules of intermittent reinforcement are also similar in laboratory animals and humans (Griffiths et al., 1980a) suggesting that the same processes are mediating the behaviors.

There are several distinct advantages to using human subjects in experimental models of drug abuse. First, the inferential steps inherent in generalizing from animal to human drug-taking behavior are avoided. Second, the relation between the subjective effects and the more overt behavioral effects of a drug can be studied in humans. It has often been assumed that certain subjective effects of drugs, such as drug-induced euphoria, are associated with, and may even underlie, the behavior of drug-seeking. This assumption has led some researchers to use only subjective effects of pharmacological agents in humans to predict dependence potential of new, unknown drugs (e.g., Martin & Fraser, 1961). Although there is reason to expect a close relationship between these measures, the association between drug-induced euphoria or any other subjective effect and the behavior of drug-taking must be examined empirically (see Henningfield, Johnson, & Jasinski, this volume; Schuster, Fischman, & Johanson, 1981). It must first be shown that systematic relations exist between drug-induced mood changes and the reinforcing effects of a drug. The further question of whether there is a causal relation between the two actions can only be inferred from the degree of covariance between them under different conditions. Among the possible relations that may exist between subjective effects and behavioral effects of drugs are, for example, that certain drugs are euphorigenic but are not self-administered, while others may be self-administered but produce no euphoria. With yet other drugs, mood alterations other than euphoria, such as decreased anxiety or decreased depression, may form the basis of drug-seeking behavior. Finally, in certain cases the mood-altering effects of a drug may bear no systematic relation at all to whether individuals take it or not. These important relations can only be addressed in behavioral experiments using humans subjects who can report on their internal states.

The use of humans as subjects is also advantageous in the study of individual differences in responsiveness to drugs. Considerable individual differences have been noted in both humans’ and animals’ tendencies to self-administer drugs. In animal studies these inter-animal differences are often ignored, probably because of the difficulties inherent in studying them. In human subjects, however, these inter-subject differences constitute a rich source of information. A wide range of techniques are available to measure subject differences (e.g., personality tests) and these, combined with a laboratory test of the reinforcing properties of drugs, may reveal important factors underlying drug-taking behavior.

The use of humans as subjects in drug research also has limitations. One of these concerns the amount of control that is attainable over human subjects’ current lifestyles and past histories. Animal researchers can arrange that animals within a study have similar pre-experimental histories (e.g., constant supplier, controlled housing and diet conditions, controlled drug and experimental histories). While some aspects of the animals’ histories inevitably remain unknown, particularly in the case of monkeys not bred in captivity, the animals’ histories can be carefully controlled after arrival in a laboratory. Considerably less control is possible with human subjects, whose diverse experiences, cognitions and current states may influence their behavior in unknown ways, even under highly controlled experimental conditions. Although attempts are made to measure and relate some of these historical variables (e.g., drug use history) to the experimental outcome, these data are limited by their correlational nature. Paradoxically, the variability among individuals’ responses that results from the diversity of prior experiences may also lead to the discovery of variables that influence drug-taking behavior simply by increasing the range of factors under study.

Another major limitation of research with human subjects stems from ethical considerations. Clearly, many of the procedures commonly used in animal experiments (e.g., physiological interventions, high drug doses) are not possible with humans, limiting the range of experimental questions that can be addressed using this species. In addition to these procedural limitations, researchers in this area are aware that any exposure to a highly abused drug in an experimental context may entail some risk of subsequent abuse by subjects. This consideration may have practical implications in the experiments such as limiting the range of subjects that can ethically be accepted as well as limiting the frequency with which the drugs can be administered. Thus, ironically, the process of identifying risk factors in individuals may put them at slightly greater risk.

Drug Self-Administration Procedures in Humans

Experimenters have used at least two methods to measure the reinforcing properties of drugs in humans. One method uses rate of responding as the dependent variable. In studies using this approach, moderate doses of drugs are made available contingent upon performance of an operant response, such as pushing a button, usually according to a schedule of intermittent reinforcement. The rate of responding is taken as the indicator of the drug’s reinforcing efficacy, and the pattern of drug-taking over time can be studied. However, these procedures often entail repeated drug administrations whose frequencies are controlled by the subject. This makes it difficult to determine accurately the quality, magnitude and duration of the drug’s direct effects on the organism (e.g., on mood). Rate of responding as a measure of the efficacy of any type of reinforcer has limitations, and this is especially true in the case of drug reinforcers: Factors unrelated to the efficacy of the reinforcer (such as drug effects on the animal’s motoric capacity) can influence the rate of responding (Schuster & Johanson, 1975). Methods that employ repeated administrations of drugs in human subjects are also undesirable for ethical reasons.

Preference or choice procedures are a second method that has been used in assessing the reinforcing properties of drugs in humans. These procedures provide a rate-free measure of a drug’s reinforcing properties and minimize exposure to the drugs. Typically, subjects initially sample each of two substances on a small number of sampling trials and then choose whichever they prefer on several discrete choice trials. The reinforcing efficacy of a drug is determined by the number of trials it is chosen over a placebo or another drug. Exposure to the drug during sampling is minimized, and the experimenter can arrange to space the trials widely apart (even over days) so that the magnitude and time course of the drug’s effects can be accurately monitored.

Applications of Human Preference Tests

This laboratory has been testing drug preference in humans for the past 10 years (e.g., de Wit, Johanson, & Uhlenhuth, 1984; de Wit, Johanson, Uhlenhuth, & McCracken, 1983; de Wit, Uhlenhuth, Hedeker, McCracken, & Johanson, 1986a; de Wit, Uhlenhuth, & Johanson, 1985a, 1985b; Johanson & Uhlenhuth, 1977, 1978, 1980a, 1980b). The studies have addressed a variety of questions, including questions about the properties of particular pharmacological agents, about subject variables that influence responses to drugs, and about environmental circumstances that affect responses to drugs. The basic procedure used in these studies will be discussed in the next section, followed by a selection of our findings that will illustrate the applications of the procedure.



In most studies male and female volunteers aged between 21 and 35 are recruited from the university community through newspaper advertisements, posters, and word-of-mouth referrals. About 50% to 75% are students and most of the remainder are faculty or staff associated with the university and hospital. The number of subjects tested in each experiment varies from 12 to 24.

All subjects are carefully screened prior to acceptance into a study. After the initial telephone contact, a screening interview is conducted during which potential subjects complete a battery of psychological tests (e.g., personality tests and a psychiatric symptom checklist) as well as questionnaires pertaining to their medical history and their current and lifetime drug use. A complete psychiatric interview is conducted, and potential subjects with a history of psychosis, major depression, or drug abuse are not accepted. Screening for physical health includes an electrocardiogram and physical examination and, depending on the drug tested, may include further tests.

Prior to participation, subjects sign a consent form which outlines the details of the study and lists the drugs that they might receive. A list of all possible side effects is provided for any drug they might be given. The consent form stipulates that the subject agrees not to take other drugs except his/her normal amounts of coffee and cigarettes before and after taking a capsule for periods of time that are considered safe to prevent drug interactions. In addition, the subject agrees not to drive or operate heavy machinery for 6 hours after capsule ingestion.


In each experiment the subjects’ relative preference for one of two substances is measured. In most cases the substances compared are a drug and a placebo, but they can also be different doses of one drug or two different drugs. Each experiment consists of seven or nine sessions conducted over a period of 2 or 3 weeks. On each session subjects report to the laboratory between 9 and 10 a.m. to complete pre-drug mood or subjective-effects questionnaires (see below) and to ingest a capsule. Subjects are instructed to note the color of the capsule and to try to associate any drug effects with that color capsule. They are told that the same drug will always be contained in the same color capsule. After taking the capsule, subjects are free to leave the laboratory to resume their normal daily activities, taking with them additional questionnaires to be filled out 1, 3, and 6 hours later. An additional questionnaire is completed at Hour 6 on which subjects indicate how much they liked the drug, whether they noticed any unusual reactions, and what type of drug they thought they had received.

On the first four sessions of each experiment, subjects are given the two substances to sample, once per session, twice in each alternating order. The drug and placebo (or other drug) are contained in color-coded capsules to facilitate future identification, and the colors are varied among subjects to control for color preferences. On the last three to five sessions of each experiment, the procedure is identical except that the subjects are asked to choose between the two different colored capsules. The number of times one substance is chosen over another is taken as an indicator of its reinforcing properties. When the comparison is between a drug and a placebo, it is assumed that the placebo is neutral and that any consistent deviation from chance level of choices of the two substances is due to the drug’s reinforcing or aversive properties. When the comparison is between two drugs, preference for one drug may indicate either that one drug has relatively greater positive reinforcing properties or that the other has relatively greater aversive properties.

By allowing subjects to leave the laboratory immediately after taking their capsules on each session, some control over the subjects’ behavior is lost, such as in the accuracy of recording the drugs’ subjective effects. Nevertheless, the results obtained to date indicate that the method is sensitive and reliable, and that, at least with the subject population currently being tested, subjects complete their questionnaires promptly and candidly. Moreover, mood changes induced by the drugs appear to override normal fluctuations in mood that occur during the course of the day. The subjective-effects measures have proven to be sensitive to differences in the time course and dose of drugs administered and even reflect the changes in mood that occur over the course of the day on sessions when only placebo is administered (e.g., an increase in fatigue in the late afternoon). Thus, despite the relative lack of control of the subjects’ behavior (compared to strictly laboratory studies) and despite the myriad other factors that can be presumed to influence mood during the course of the day, the subjective-effects measures appear to be highly sensitive to the drugs’ effects. From a practical point of view, the ‘outpatient’ nature of the procedure entails considerable savings in time and expense for the researcher. From a theoretical point of view, it could be argued that testing the subjects in their normal daily environments increases the validity of the procedure by testing drug responses in a naturalistic setting. This issue will be discussed in greater detail in a later section.

Subjective Effects

The scales used to assess subjective effects include an experimental version of the Profile of Mood States (POMS; McNair, Lorr, & Droppelman, 1971) and a 49-item version of the Addiction Research Center Inventory (ARCI; Haertzen, 1966; Martin, Sloan, Sapira, & Jasinski, 1971; see also Haertzen & Hickey, this volume). The POMS consists of 72 adjectives commonly used to describe momentary mood states. Subjects indicate how they feel at the moment in relation to each of the adjectives on a 5-point scale ranging from not at all (0) to extremely (4). Eight clusters of items have been empirically derived using factor analysis. These clusters, which form the eight subscales of the questionnaire, were given names that best described the clustered adjectives: Anxiety, Depression, Anger, Vigor, Fatigue, Confusion, Friendliness and Elation. Two additional subscales were derived on an intuitive basis from other scales: Arousal = (Anxiety + Vigor) - (Fatigue + Confusion) and Positive Mood = Elation - Depression. Although the POMS was originally developed for use with only a single-administration, researchers in this laboratory pioneered its use with repeated administrations to monitor mood changes over time.

The shortened form of the ARCI consists of 49 true/false items which have been separated into five clusters described as measuring typical drug effects. The Amphetamine (A) and Benzedrine Group (BG) scales measure stimulant-like effects, the Morphine-Benzedrine Group (MBG) scale is thought to reflect drug-induced euphoria, the Pentobarbital-Chlorpromazine-Alcohol Group (PCAG) scale measures sedative-like effects, and the Lysergic-Acid (LSD) scale reflects psychotomimetic effects, often also described as dysphoria (see Haertzen & Hickey, this volume).

Data from these subjective-effects questionnaires can be analyzed using any one of a number of statistical techniques. Most commonly, two- or three-way repeated measures analyses of variance are used, with the factors being Drug (i.e., drugs versus placebo), Hour, and, when comparisons are made among different subject groups, a subject grouping factor (e.g., drug-choosers versus drug non-choosers). The analyses are performed separately on each scale, using the subjects’ average scores at each hour on drug and placebo sessions. Data from the sampling sessions only (Sessions 1 to 4) are utilized to minimize expectancy effects which might occur on choice sessions. When significant (p < 0.05) Drug-By-Hour interactions are obtained on any of the subscales, post-hoc tests are conducted to determine the hours at which the drug and placebo scores differ. The data can also be analyzed in other ways. For example, multivariate methods may be used to evaluate a drug’s relative effects on different dependence measures (e.g., on different subscales of the mood questionnaires). Analyses may also be performed on difference scores, differences either between pre- and post-drug scores or between placebo and drug session scores. Trend analyses are often useful to describe the time course or dose dependence of a drug’s effects. Selection of the appropriate method of analysis depends on the goals of the study and the nature of the data.

Experimental Findings

The choice procedure has been used to evaluate the reinforcing properties of a range of drugs, including stimulants and anorectics as well as tranquilizers and sedatives. The method was first tested with a well-known drug of abuse, d-amphetamine (Johanson & Uhlenhuth, 1980a). Amphetamine is a drug that is reliably self-administered by laboratory animals (Pickens & Harris, 1968). Relative to other drugs, it is a robust reinforcer in many species and under varied environmental circumstances (Johanson & Schuster, 1981). It was therefore not surprising that the majority of human subjects tested in the choice test chose amphetamine (5 mg d-amphetamine) more often than placebo. Subjects (N = 31) chose the drug an average 4 of the 5 choice opportunities. Moreover, amphetamine produced typical, stimulant-like subjective effects and subjects verbally reported liking the drug. Relative to placebo session scores, amphetamine session scores were higher for the Vigor, Elation, Friendliness, Arousal, and Positive Mood subscales of the POMS and lower for the Confusion scale. In subsequent experiments similar results have been obtained with amphetamine and several other stimulant-like drugs. The results of these experiments are summarized in Table 1. Anorectic drugs that produced subjective effects similar to those of amphetamine (e.g., benzphetamine and phenmetrazine) were, on the average, chosen more often than placebo, whereas drugs that produced distinctively different subjective effects (e.g., mazindol and fenfluramine) were chosen as often or less often than placebo.

The procedure has also been used to evaluate the reinforcing properties of the widely prescribed tranquilizer, diazepam (Valium; de Wit et al., 1983, 1986a, 1986b; Johanson & Uhlenhuth, 1980b). When tested in laboratory animals, this and other benzodiazepine drugs have only minimal reinforcing properties compared to other classes of drugs such as stimulants and opiates. There have been only occasional reports of animals self-administering benzodiazepines (Bergman & Johanson, 1985; Griffiths & Ator, 1981; Weeks & Collins, this volume; cf. Yanagita, this volume). The extent of abuse of these drugs in humans has been the subject of considerable controversy (Dietch, 1983; Greenblatt, Shader, & Abernethy, 1983; Rickels, 1981; Woody, O’Brien, & Greenstein, 1975). While there is concern about the large number of prescriptions that are written for these drugs, there have been relatively few documented cases of misuse of these drugs compared to other well-known drugs of abuse (Marks, 1978). Although physical dependence has been reported in some cases after prolonged exposure to therapeutic doses of tranquilizers such as diazepam, the incidence of abuse of benzodiazepines is strikingly low in view of the large number of people who are exposed to these drugs through prescription use (Mellinger, Balter, & Uhlenhuth, 1984). Preference for diazepam has been tested in this laboratory in several studies. Normal volunteers chose placebo either as often as diazepam or, at higher doses, showed an avoidance of the drug (de Wit et al., 1983; Johanson & Uhlenhuth, 1980b). Whereas lower doses produced negligible subjective effects, the higher doses produced sedation (e.g., decreased Vigor and Arousal and increased Confusion and Fatigue subscale scores of the POMS). Similar results have been obtained with two other benzodiazepines, lorazepam and flurazepam (de Wit et al., 1984, 1985b). Doses of any of these drugs that produced noticeable subjective effects were avoided in the preference test by most of the subjects.

Table 1
Percent Drug Choice in Experiments With Several Amphetamine-Like and Non-Amphetamine-Like Anorectic Drugs
Drug Doses Tested Percent Choice*  Reference**
A. Drugs with amphetamine-like subjective effects
d-amphetamine (N = 24) 
benzphetamine (N = 11) 
phenmetrazine (N = 13) 
diethylpropion (N = 10)
5, 10 mg 
25, 50 mg 
25, 50 mg 
25 mg

B. Drugs with partially amphetamine-like subjective effects
(N = 12, 17) 
caffeine (N = 18)
12.5, 25, 50, 75 mg 

100, 300 mg




C. Drugs without amphetamine-like subjective effects
mazindol (N = 12) 
fenfluramine (N = 14)
0.5, 1.0, 2.0 mg 
20 mg



* Percent drug choice is shown for the dose of drug (underlined) at which the highest level of choice was obtained. 

**  (references) 
1: de Wit, unpublished observations 
2: Chait, unpublished observations 
3: Johanson & Uhlenhuth, 1978 
4: Johanson & Uhlenhuth, 1982


Preferences for several other drugs with sedative properties (e.g., alcohol, pentobarbital) have also been tested using a modified choice procedure in which subjects are tested in a controlled laboratory environment. Results of these studies will be discussed in a later section.

Subject Selection

The results of drug preference tests depend on the subjects who are tested. The appropriateness of a particular subject population in any study depends in part on the experimental question. For example, it has been found subjects with a history of drug abuse are more likely than non-abusing subjects to prefer a drug (such as diazepam) with equivocal reinforcing properties (Griffiths, Bigelow, Liebson, & Kaliszak, 1980b). This may be especially true if the drug shares certain properties with the drugs that these individuals have abused. Thus, these subjects may be appropriate for testing the dependence potential of a new therapeutic agent, particularly if this agent is likely to be prescribed to them and/or if it is believed to have some properties in common with the subjects’ known drugs of abuse. However, because of their complex drug use histories, these subjects may be less appropriate in studies designed to assess risks for abuse in the general population or to study individual differences in risk for abuse among drug-naive individuals. For example, our research has shown that normal volunteers do not prefer the tranquilizer diazepam over a placebo, and in fact they avoid it at doses that produced any appreciable subjective effects. This is consistent with epidemiological data from other sources (Mellinger et al., 1984) indicating that the large majority of individuals exposed to benzodiazepines therapeutically or recreationally do not develop patterns of excessive use or abuse of these drugs. Nevertheless, the possibility that some individuals are at risk for dependence on benzodiazepines forms the basis of our investigations with non-drug-abusing subjects.

The drug preference studies in our laboratory have focused on both the behavioral and the subjective responses of non-drug-abusing individuals to identify possible risk factors for abuse. Some studies have utilized a heterogeneous sample of subjects, screened only for age limitations and health for safety reasons, whereas other studies have used individuals who are hypothesized to be a higher-than-average risk for abuse because they possess certain characteristics. The first method may give an indication of the likelihood of abuse in a relatively general or unselected population (taking into account the limitations due to sampling a particular geographic or demographic group and due to the voluntary basis of participation in any experimental study). An advantage of a relatively unselected population is that it provides the variability necessary to study individual differences in responses to the drugs. We have conducted several post-hoc analyses of drug preference studies to explore subject characteristics associated with preferences for or aversions to certain drugs (de Wit, McCracken, Uhlenhuth, & Johanson, 1987a; de Wit, Uhlenhuth, & Johanson, 1986b; de Wit, Uhlenhuth, Pierri, & Johanson, 1987b; Uhlenhuth, Johanson, Kilgore, & Kobasa, 1981). These analyses have revealed interesting individual differences in subjective response to drugs such as alcohol and amphetamine. For example, subjects who report primarily stimulant-like subjective effects after an acute dose of alcohol (0.5 g/kg) were more likely to show behavioral preference for this drug than subjects who reported predominantly sedative effects from the drug (de Wit et al., 1987b). In another experiment in which preference for amphetamine over placebo was measured, the few individuals who did not prefer the amphetamine in the choice test reported increases in measures of anxiety and depression and no typical stimulant-like effects after ingestion of the drug during sampling sessions (de Wit et al., 1986b). Intuitively logical relations were found in these experiments between the subjective effects of drugs and subjects’ behavioral preference (choice). Comparisons can also be made across different drugs to study how drug preferences and aversions can be associated with different subjective effects and with different subject characteristics, depending on the drug.

Other studies have utilized special subject populations hypothesized to be at-risk for abuse. These studies have focused on the "self-medications hypothesis" or the notion that a drug can acquire reinforcing properties through its ability to relieve an aversive state. One subject characteristic hypothesized to be associated with preference for diazepam is anxiety level. The efficacy of benzodiazepines as anxiety-reducing agents is well-known. Thus, assuming that the state of anxiety is an aversive condition, the relief of this state with a drug such as diazepam may be reinforcing in highly anxious subjects. We tested this hypothesis in three groups of anxious subjects (de Wit et al., 1986a, 1987a). One group consisted of subjects who scored high on two standardized anxiety tests (see Speilberger, Gorsuch, & Lushene, 1970; Taylor, 1953) but did not meet criteria for a psychiatric diagnosis. Two groups met criteria for Generalized Anxiety Disorder (APA, 1980). One of these groups consisted of subjects who were sufficiently distressed by their anxiety to desire treatment, whereas the other group was not actively seeking treatment. A control group consisted of subjects who scored within the normal range on anxiety tests. The percent of diazepam choice in each of these groups is summarized in Table 2. The symptomatic subjects did not choose diazepam more often, on the average, than the control subjects. It was found, however, that a small group (N = 4) of the anxious subjects seeking treatment did derive positive effects (i.e., elevated liking scores and apparently positive subjective effects) from the drug and consistently chose it over placebo. Whether these subjects should be considered at-risk for abuse of diazepam or whether their preference in the experiment reflects the genuine therapeutic efficacy of the drug is a question that may be addressed in future research. As important as the finding that certain highly selected subjects did prefer the drug, however, is the finding that the majority of even highly anxious subjects did not choose the drug in preference to a placebo, despite its measurable anxiolytic effects. These findings suggest that the risk for abuse, even for individuals who are most likely to be prescribed these drugs, is low.

Another study examined diazepam preference in older adults. Epidemiological data show that the prescription use of benzodiazepines increases with patients’ age (Mellinger et al., 1984). Because high prescription rates are sometimes indicators of excessive or nontherapeutic drug use, the possibility that the reinforcing properties of diazepam increase with age was examined. It was found, however, that subjects aged 40 to 55 years did not exhibit a higher preference for 10 mg diazepam than the control group aged 21 to 35 years.

Table 2
Percent Diazepam Choice in Different Subject Populations
Subject Description Percent Drug Choice Reference**

(5 mg) (10 mg)
Normal control (N = 9) 
Normal control (N = 12) 
Anxious mood (N = 11) 
Anxiety disorder (N = 12) 
Anxiety disorder, seeking treatment (N = 14) 
Older* (N = 11)



* Age 40 to 55 (All remaining subjects were 21 to 35 years of age.) 

**  (references) 
1: Johanson & Uhlenhuth, 1980a 
2: de Wit et al., 1986a 
3: de Wit et al., 1987b 
4: de Wit et al., 1986b


Future research of this type may yet reveal subject characteristics which are associated with strong preferences for benzodiazepines or other sedative drugs and allow us to identify patterns of responses that are predictive of risk for abuse. The results with the large majority of subjects tested to date, however, are consistent with epidemiological data and animal self-administration data indicating that, relative to other abused drugs, the risk for dependence on these drugs is low.

The "self-medication hypothesis" has also been investigated in special subject groups with the stimulant amphetamine. In a recent study amphetamine preference was tested in depressed subjects (because of the drug’s purported antidepressant properties) and in subjects concerned about being overweight (because of its anorectic properties). We found that neither of these populations showed a stronger preference for amphetamine than a control group (de Wit, unpublished observations).

Thus, the choice procedure can be used with different subject groups to study individual differences in subjective and behavioral responses to drug in a non-abusing population and to test specific hypotheses concerning the basis of non-medical drug use.

Environmental Conditions

Experimental research in behavioral pharmacology over the last 20 years has provided abundant evidence of the importance of environmental variables affecting behavior maintained by drugs as reinforcers. It has been shown that both conditioned and unconditioned contextual factors can increase and decrease responding for drugs across a wide range of drugs, species, and individuals (Johanson, 1975; Young, Herling, & Woods, 1981). In view of these strong environmental influences, results from individual experiments must be interpreted with caution, since they generally employ only one, closely circumscribed set of experimental conditions.

The context in which a drug’s effects are experienced may influence whether it is preferred in a choice test. For example, when stimulant drugs are tested in the daytime, outpatient procedure, their subjective effects are consistent with and may even facilitate the subjects’ daily activities requiring concentration and alertness. In contrast, sedative-like drugs, some of which may have positive subjective effects in other contexts, are likely to be avoided in the daytime test because they interfere with psychomotor and cognitive performance. Therefore, we adapted our methodology to test preferences for sedative-like drugs in a laboratory-based recreational environment, where outside demands on the subjects’ behavior and attention were eliminated. Subjects are tested during the evenings in a comfortable room with a couch, television, games, and movies. They are tested in groups of four and are free to engage in recreational activities of their choice. Under these circumstances preliminary results indicate that a substantial proportion of subjects will choose drugs with sedative properties, such as alcohol (50% to 75% drug choice, depending on experimental conditions) and pentobarbital (45% to 52% drug choice). The level of drug choice appears to be related to both subject characteristics (e.g., drinking history) and aspects of the testing environment (e.g., dose regimen). Preference studies with diazepam under these circumstances are currently in progress. Although direct comparisons of results obtained in the outpatient and the laboratory procedures cannot be made, the preliminary findings obtained with ethanol and pentobarbital are consistent with what is expected from other laboratory and epidemiological data. Further data are needed to establish whether the laboratory procedure will provide a valid measure of drug preferences in normal volunteers. It seems likely, however, that a suitable, undemanding context in which to experience the drugs’ effects may be a necessary condition for the observation of reinforcement from sedative-like drugs.

Environmental variables have also been shown to attenuate the reinforcing properties of an otherwise highly reinforcing drug, amphetamine. Amphetamine is preferred over placebo when it is administered during the course of normal working days (de Wit et al., 1986b; Johanson & Uhlenhuth, 1980a). However, subjects who were required to take their capsules late in the day (between 4 and 5 p.m.) showed a lower preference (53% drug choice) for 5 mg d,l-amphetamine over placebo than subjects who took the drug between 9 and 10 a.m. (80% drug choice; de Wit et al., 1985a). Some subjects in the afternoon group complained that the drug interfered with their ability to sleep at the normal times, which may have been a reason for the decreased preference. Whatever the reasons, this experiment indicated that a nonpharmacological variable such as time of day of drug administration can influence the drugs efficacy as a reinforcer. Other researchers (Griffiths, Bigelow, & Liebson, 1977) have also shown that self-administration of a drug (i.e., alcohol) which is normally reinforcing in humans can be decreased by an environmental manipulation such as a time-out procedure.


In this chapter we have described a drug preference procedure that has been used to test the reinforcing properties of drugs in humans. Some of the advantages as well as the problems with the use of human experimental subjects were discussed, and these were illustrated with examples from this laboratory. The method can be used to address experimental questions about the pharmacological properties of an unknown drug or about organismic or environmental variables that influence the reinforcing properties of drugs in humans. The method also provides an opportunity to integrate two historically independent approaches to the study of drug dependence, human subjective-effects reports and animal self-administration techniques. Whereas the subjective-effects measures in humans have traditionally been obtained without a concurrent behavioral measure of reinforcement, the animals self-administration experiments have been limited to the measurement of overt behavior. An exploration of the relationship between the internal stimulus properties of a drug and its effects on behavior, in particular its reinforcing effects, using procedures such as the one described here will improve our understanding of the determinants of excessive drug use in humans.


Our thanks to M. Fischman, G. Heyman, and L. Chait for their helpful comments. The research was supported by a grant from the National Institute on Drug Abuse (DA 02812).


American Psychiatric Association. (1980). Diagnostic and statistical manual of mental disorders, third edition. American Psychiatric Association Washington, DC.

Bergman, J., & Johanson, C. E. (1985). The reinforcing properties of diazepam under several conditions in the rhesus monkey. Psychopharmacology, 86, 108-113.

de Wit, H., Johanson, C. E., & Uhlenhuth, E. H. (1984). Reinforcing properties of lorazepam in normal volunteers. Drug and Alcohol Dependence, 13, 31-41.

de Wit, H., Johanson, C. E., Uhlenhuth, E. H., & McCracken, S. (1983). The effects of two non-pharmacological variables on drug preference in humans. In L. S. Harris (Ed.), Problems of drug dependence, 1982 (National Institute on Drug Abuse Research Monograph 43, pp. 251-257). Washington, DC: U.S. Government Printing Office.

de Wit, H., McCracken, S. M., Uhlenhuth, E. H., & Johanson, C. E. (1987a). Diazepam preference in subjects seeking treatment for anxiety. In L. S. Harris (Ed.), Problems of drug dependence, 1986 (National Institute on Drug Abuse Research Monograph 76, pp. 248-254). Washington, DC: U.S. Government Printing Office.

de Wit, H., Uhlenhuth, E. H., Hedeker, D., McCracken, S., & Johanson, C. E. (1986a). Lack of preference for diazepam in anxious volunteers. Archives of General Psychiatry, 43, 533-541.

de Wit, H., Uhlenhuth, E. H., & Johanson, C. E. (1985a). Drug preference in normal volunteers: Effects of age and time of day. Psychopharmacology, 87, 186-193.

de Wit, H., Uhlenhuth, E. H., & Johanson, C. E. (1985b). Lack of preference for flurazepam in normal volunteers. Pharmacology Biochemistry & Behavior, 21, 865-869.

de Wit, H., Uhlenhuth, E. H., & Johanson, C. E. (1986b). Individual differences in the behavioral and subjective effects of amphetamine and diazepam. Drug and Alcohol Dependence, 16, 341-360.

de Wit, H., Uhlenhuth, E. H., Pierri, J., & Johanson, C. E. (1987b). Individual differences in behavioral and subjective response to alcohol. Alcoholism: Clinical and Experimental Research, in press.

Dietch, J. (1983). The nature and extent of benzodiazepine abuse: An overview of recent literature. Hospital and Community Psychiatry, 34, 1139-1145.

Greenblatt, D. J., Shader, R. L., & Abernethy, D. R. (1983). Current status of benzodiazepines, Parts 1 and 2. New England Journal of Medicine, 309, 354-358 & 410-416.

Griffiths, R. R., & Ator, N. A. (1981). Benzodiazepine self-administration in animals and humans: A comprehensive literature review. In S. I. Szara & J. P. Ludford, (Eds.), Benzodiazepines: A review of research results, 1980 (National Institute on Drug Abuse Research Monograph 33, pp. 22-36). Washington, DC: U.S. Government Printing Office.

Griffiths, R. R., Bigelow, G. E., & Henningfield, J. E. (1980a). Similarities in human and animal drug-taking behavior. In N. K. Mello (Ed.), Advances in substance abuse: Behavioral and biological research (Vol. I, pp. 1-90). Greenwich, CT: JAI Press.

Griffiths, R. R., Bigelow, G., & Liebson, I. (1976). Human sedative self-administration: Effects of interingestion interval and dose. Journal of Pharmacology and Experimental Therapeutics, 197, 488-494.

Griffiths, R. R., Bigelow, G., & Liebson, I. (1977). Comparison of social time-out and activity time-out procedures in suppressing ethanol self-administration in alcoholics. Behavioral Research and Therapy, 15, 329-336.

Griffiths, R. R., Bigelow, G. E., Liebson, I., & Kaliszak, J. E. (1980b). Drug preference in humans: Double-blind choice comparison of pentobarbital, diazepam, chlorpromazine and placebo. Journal of Pharmacology and Experimental Therapeutics, 215, 649-661.

Haertzen, C. A. (1966). Development of scales based on patterns of drug effects using the Addiction Research Center Inventory. Psychological Reports, 18, 163-194.

Johanson, C. E. (1975). Pharmacological and environmental variables affecting drug preference in rhesus monkeys. Pharmacological Review, 27, 343-355.

Johanson, C. E., & Balster, R. L. (1978). A summary of the results of a drug self-administration study using substitution procedures in primates. Bulletin of the Narcotics, 30, 43-54.

Johanson, C. E., & Schuster, C. R. (1981). Animal models of self-administration. In N. K. Mello (Ed.), Advances in substance abuse: Behavioral and biological research, Vol. 2. Connecticut: JAI Press, 219-297.

Johanson, C. E., & Uhlenhuth, E. H. (1977). Drug preference in humans. Paper presented at the 1977 meeting of the American Psychological Association, San Francisco.

Johanson, C. E., & Uhlenhuth, E. H. (1978). Drug self-administration in humans. In N. A. Krasnegor (Ed.), Self-administration of abused substances: Methods of study (National Institute on Drug Abuse Research Monograph 20, pp. 68-85). Washington, DC: U.S. Government Printing Office.

Johanson, C. E., & Uhlenhuth, E. H. (1980a). Drug preference and mood in humans: D-amphetamine. Psychopharmacology, 72, 275-279.

Johanson, C. E., & Uhlenhuth, E. H. (1980b). Drug preference and mood in humans: Diazepam. Psychopharmacology, 71, 269-273.

Johanson, C. E., & Uhlenhuth, E. H. (1982). Drug preference in humans. Federation Proceedings, 233, 228-233.

Marks, J. (Ed.) (1978). The benzodiazepines. Baltimore: University Park Press.Martin W. R., & Fraser, H. F. (1961). A comparative study of physiological and subjective effects of heroin and morphine administered intravenously in postaddicts. Journal of Pharmacology and Experimental Therapeutics, 133, 388-399.

Martin, W. R., Sloan, J. W., Sapira, J. D., & Jasinski, D. R. (1971). Physiologic, subjective and behavioral effects of amphetamine, methamphetamine, ephedrine, phenmetrazine and methylphenidate in man. Clinical Pharmacology and Therapeutics, 12, 245-258.

McNair, D. M., Lorr, M., & Droppelman, L. F. (1971). Profile of mood states (Manual). San Diego, CA: Educational and Industrial Testing Service.

Mellinger, G. D., Balter, M. B., & Uhlenhuth, E. H. (1984). Prevalence and correlates of the long-term use of anxiolytics. Journal of the American Medical Association, 251, 375-379.

Mello, N. K., & Mendelson, J. H. (1971). A quantitative analysis of drinking patterns in alcoholics. Archives of General Psychiatry, 25, 527-539.

Pickens, R., & Harris, W. C. (1968). Self-administration of d-amphetamine in rats: Effects of reinforcement magnitude and fixed-ratio size. Psychopharmacologia, 12, 158-163.

Rickels, K. (1981). Are benzodiazepines overused and abused? British Journal of Clinical Pharmacology, 11, 715-835.

Schuster, C. R., Fischman, M. W., & Johanson, C. E. (1981). Internal stimulus control and subjective effects of drugs. In T. Thompson & C. E. Johanson (Eds.), Behavioral pharmacology of human drug dependence (National Institute on Drug Abuse Research Monograph 37, pp. 116-129). Washington, DC: U.S. Government Printing Office.

Schuster, C. R., & Johanson, C. E. (1975). The use of animal models for the study of drug abuse. In R. J. Gibbins, Y. Israel, H. Kalant, R. E. Popham, R. E. Schmidt, and R. B. Smart, (Eds.), Recent advances in alcohol and drug problems (Vol. 1, pp. 1-31). New York: John Wiley and Sons.

Speilberger, C. E., Gorsuch, R. L., & Lushene, R. E. (1970). State-trait anxiety inventory. Test manual for form X. Palo Alto, CA: Consulting Psychologists Press.

Taylor, J. A. (1953). A personality scale of manifest anxiety. Journal of Abnormal and Social Psychology, 48, 285-290.

Thompson, T., & K. R. Unna (Eds.) (1977). Predicting dependence liability of stimulant and depressant drugs. Baltimore: University Park Press.

Uhlenhuth, E. H., Johanson, C. E., Kilgore, K., & Kobasa, S. C. (1981). Drug preference and mood in humans: Preference for d-amphetamine and subject characteristics. Psychopharmacology, 74, 191-194.

Woody, G. E., O’Brien, C. P., & Greenstein, R. (1975). Misuse and abuse of diazepam: An increasing common medical problem. International Journal of the Addictions, 10, 843-848.

Young, A. M., Herling, S., & Woods, J. H. (1981). History of drug exposure as a determinant of drug self-administration. In T. Thompson & C. E. Johanson (Eds.), Behavioral pharmacology of human drug dependence (National Institute on Drug Abuse Research Monograph 37, pp. 75-89). Washington, DC: U.S. Government Printing Office.

©1987 Springer-Verlag (printed version)
©2000-2009 Addiction Science Network (web-enhanced version)

ASNet Home

ASNet Profile

Illicit Drug Index

Virtual Lab Tour

Research Reports

Drug Classification

A Primer on Drug Addiction

Experimental Methods

Treatment Resources

Biological Basis of Addiction

Before Prohibition: Early Psychoactive Medicines

Links to Other Websites

Click here to enter the Addiction Science Network Discussion Forum

Brain Reward System©1999-2009 Addiction Science Network
This page was last revised 06 April 2009 12:36 EDT.
Send comments to:
Report technical problems to: