1991, 56, 377-393

JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR

NUMBER

2

(SEPrEMBER)

BEHAVIORAL ECONOMICS OF DRUG SELF-ADMINISTRATION AND DRUG ABUSE POLICY STEVEN R. HURSH WALTER REED ARMY INSTITUTE OF RESEARCH The concepts of behavioral economics have proven useful for understanding the environmental control of overall levels of responding for a variety of commodities, including reinforcement by drug selfadministration. These general concepts are summarized for application to the analysis of drug-reinforced behavior and proposed as the basis for future applications. This behavioral agenda includes the assessment of abuse liability, the assay of drug-reinforcer interactions, the design of drug abuse interventions, and the formulation of drug abuse public policy. These separate domains of investigation are described as part of an overall strategy for designing model projects to control drug use and testing public policy initiatives. Key words: behavioral economics, demand curve, elasticity, drug self-administration, cost, unit price, FR schedule, public policy, design of culture

Concepts of behavioral economics have proven useful for understanding the environmental control of overall levels of responding for a variety of commodities in closed systems (Foltin, in press; Hursh, 1984; Lea, 1978; Lea & Roper, 1977; Rashotte & Henderson, 1988). Recently, several publications have demonstrated the applicability of these same principles for describing the strength of performances maintained by drug reinforcement (Bickel, DeGrandpre, Higgins, & Hughes, 1990; Bickel, DeGrandpre, Hughes, & Higgins, 1991). This paper summarizes the basic principles of behavioral economics as applied to drug-reinforced behavior and proposes the following agenda for future applications: (a) assessment of abuse liability, (b) assay of drugreinforcer interactions, (c) framework for designing drug abuse interventions, and (d) strategy for formulating drug abuse public policy. ASSESSMENT OF ABUSE LIABILITY A large variety of drugs have been shown to serve as reinforcers for operant behavior The views of the authors do not purport to reflect the position of the Department of the Army or the Department of Defense (para 4-3, AR 360-5). The research described in this report was conducted in compliance with the Animal Welfare Act and other Federal statutes and regulations relating to animals and experiments involving animals and adheres to the principles stated in the Guide for the Care and Use of Laboratory Animals, NIH publication 86-23, 1985 edition. The author wishes to acknowledge the assistance of Raymond Genovese, Frederick Manning, Harold Black, and Gregory Galbicka during preparation of earlier drafts of the manuscript. Reprints may be obtained by writing Steven R. Hursh, Division of Neuropsychiatry, Walter Reed Army Institute of Research, Washington, D.C. 20307-5100.

(Brady & Lukas, 1984; Johanson, 1978). In general, drugs that maintain recreational use in humans will serve as reinforcers for behavior in nonhumans, the one general exception being hallucinogens, which follow an atypical pattern of use in humans as well (Griffiths, Bigelow, & Henningfield, 1980). In a typical experiment with heroin as the reinforcer (Elsmore, personal communication), monkeys were prepared with indwelling catheters for the delivery of small-volume infusions of heroin or vehicle dependent on the occurrence of keypress responses. A similar key could provide 750 mg food pellets. Without special shaping procedures, responding for heroin gradually increased to stable levels above that observed when responses produced vehicle (saline solution). Over subsequent conditions, the number of responses required to obtain an injection (fixed-ratio schedule, FR) was increased from 1 to 240. Figure 1 (bottom panel) shows the number of injections obtained per day as a function of the size of the FR. This plot of daily drug consumption as a function of the FR (price) is termed a demand curve. When plotted in log-log coordinates, the slope of the function indicates the sensitivity of consumption to changes in price. If the proportional change in consumption is less than the proportional change in price, then the slope is between 0 and -1 and is termed inelastic; if changes in consumption are proportionately greater than changes in price, then the slope is more negative than -1 and is termed elastic. For the demand curves in Figure 1, the slope was less negative than -1, and the corresponding response rates that produced these levels

377

STEVEN R. HURSH

378 ; 300 Wa a

0 0

%.100I

A Stu

A

i HEOI

EL WOOD £

100

IO/

300

z 100

z z

0 Percy

10

110 1

3 10 30 100 300 FIXED RATIO (LEVER PRESSES/REINFORCEMENT)

Fig. 1. Top panel: For 2 rhesus monkeys, daily consumption of food pellets as a function of FR schedule, in log-log coordinates. Bottom panel: For 4 rhesus monkeys, daily infusions of heroin as a function of FR schedule, in log-log coordinates.

100

1000

FIXED RATIO (LEVER PRESSES/REINFORCEMENT)

Fig. 2. For 4 rhesus monkeys, daily lever presses for food or heroin, as indicated, as a function of FR schedule.

dinates, then elasticity may be read as the point slope of the demand curve. A variety of environmental and procedural factors contribute to the level and elasticity of demand. The level of consumption show an increase with increas- of demand is controlled in part by the size of ing FR (price) (Figure 2). For comparison, 2 each "package" of the commodity (the scale of other monkeys were tested under a similar se- measurement of consumption), by characterries of FR schedules with food reinforcement, istics of the consumer (such as body weight for and the results were similarly plotted in Fig- food reinforcers), and by the current state of ures 1 (top panel) and 2 (dashed lines). For deprivation. The elasticity of demand is largely food reinforcement, demand sloped downward determined by the nature of the commodity at a less negative slope than demand for heroin ("necessity" or "luxury" good) and the avail(i.e., demand for food was more inelastic) and responding (Figure 2) increased at a greater rate with increases in FR value. Thus, demand LEVEL SHIFT (across this range of FR values) was inelastic for both commodities, but was more inelastic for food. Demand curves for a variety of commodities, including drug reinforcers, have been evaluENI ated in the laboratory (see Bickel et al., 1990, 1991; Hursh, 1980, 1984). Demand curves may be distinguished by two general charac- a teristics, illustrated in Figure 3. A demand ELASTICITY,SHIFT \ curve depicts the total consumption of a commodity as a function of price (the cost per unit of reinforcement, see unit price below). The level of demand is the total amount of consumption at minimal price, usually FR 1; the Log Price elasticity of demand is the rate of change in 3. Diagram of a hypothetical demand curve showconsumption with increasing price (see Hursh, ingFig. daily consumption as a function of price, in log-log 1980, 1984, for further explanation). If the coordinates. Also shown are two prototypical changes in demand curve is plotted in logarithmic coor- demand, a level shift and an elasticity shift. -l

BEHA VIORAL ECONOMICS OF DRUG ABUSE

379

RESPONSE OUTPUT

DEMAND




ui

00o

1000

cL

z

a: LL

cc 1000

1

00

1

tIN

100

1000

10

100

1000

FIXED RATIO FIXED RATIO Fig. 4. Left panel: Daily consumption of food or saccharin as a function of FR schedule, in log-log coordinates. Right panel: Total daily lever presses for either food or saccharin as a function of FR schedule, in log-log coordinates. Data from a representative rhesus monkey.

ability of substitutes or other sources of the commodity. Figure 4 (left panel) illustrates demand by a monkey for food and saccharinsweetened water; level of demand at a low price was adjusted to be similar by varying the size of the saccharin reinforcer. As price increased, consumption of saccharin decreased much more rapidly than consumption of food, indicating that elasticity of demand for saccharin, a nonnutritive commodity, was greater than elasticity of demand for food, a biological necessity. The right panel of Figure 4 shows the total output of responses (key presses) required to produce these reinforcers across the range of prices. These inverted U-shaped functions are typical for most commodities studied in a closed economy (Hursh, 1980, 1984; Hursh & Bauman, 1987). The level of responding for food was about 10 times that for saccharin at its peak and reached its peak at a much higher price.

Unit price. When assessing demand for a commodity across different test situations, it is important to insure that the same units of price and consumption are used for the analysis. Price is best understood as a cost-benefit ratio that describes the amount and effort of work required for each unit of reinforcement. This ratio is defined as unit price: Unit price Responses per reinforcer x Effort Size of reinforcer

Hursh, Raslear, Shurtleff, Bauman, and Simmons (1988) demonstrated that this measure was capable of providing a unitary demand curve from eight different test conditions all

scaled in unit-price terms. When using drugs as reinforcers, unit price may be defined as follows (see Bickel et al., 1990): Unit price Responses per drug reinforcer x Effort Drug dose

Likewise, consumption across conditions should be measured in comparable units across conditions, usually total drug consumption per day (frequency of infusion x dose or total volume x concentration). The application of this methodology with drug reinforcers is illustrated in a study by Lemaire and Meisch (1985). The reinforcer was a liquid mixture of pentobarbital and ethanol. Reinforcer size was studied by varying the number of liquid deliveries per period of reinforcement (e.g., 2, 8, or 16 deliveries). Combined with these sizes were different FR schedules that ranged from FR 2 to FR 256. Figure 5 shows the average demand for this drug mixture plotted as a function of unit price. For example, consumption at a unit price of 16 was observed under three combinations of FR and reinforcer size: FR 32 for 2 deliveries, FR 128 for 8 deliveries, and FR 256 for 16 deliveries. Regardless of the combination, de-

380

STEVEN R. HURSH

mand was similar when plotted in common units of price. The daily output of responses that generated this demand is shown in Figure 5. Response output increased up to a unit price of 32 and decreased at higher prices; that is, demand was initially inelastic and became elastic at unit prices above 32. FR 32 is defined as Pm.. the price that produced maximum output and demarcates the boundary from inelastic to elastic demand. This general pattern of initial inelasticity followed by elasticity at higher prices is a characteristic of most reinforcers. Bickel et al. (1990) analyzed a variety of drug self-administration studies in terms of unit-price demand curves and found this same pattern for a variety of drug reinforcers. Demand curve analysis. The smooth curve drawn through the points in Figure 5 is from an equation for demand that was fit to the data set (see Hursh et al., 1988; Hursh, Raslear, Bauman, & Black, 1989). This equation has three parameters: L for initial level of demand at minimal price, b for initial slope of the demand curve at minimal price, and a for the acceleration or increase in slope of the demand curve with increases in price. The equation is as follows, stated in the usual logarithmic units of price (P) and consumption (Q):1,2 ln(Q) = ln(L) + b(ln P) - a(P). (1) Elasticity of demand is the point slope (first derivative) of this function and is a linear function of price: Elasticity = b - a (P). Price yielding maximal output, Pma,s is: P.. = (1 + b)/a. In most cases, the b parameter is negative and close to zero so that elasticity differences are manifest in changes in a. Level shifts are I Equation 1 is derived from an exponential expression so that natural logs are required for accurate parameter estimates. The equation may be fit to demand curve data on any IBM-compatible computer using GraphPADs Inplot (GraphPAD Software, 10855 Sorrento Valley Road, #9, San Diego, California 92121). The program fits Equation 1 to the data, estimates the three parameters (+ standard errors) and the value of r2, and graphs the results (see Figure 5). Consult the author for details. 2 In an earlier paper, Hursh and Bauman (1987) proposed a different demand equation derived from economic

utility theory. Subsequent work has shown this equation to be less accurate in fitting actual data; also, it did not

yield simple expressions for elasticity and P,,,,..

seen as changes in the L parameter. For the data in Figure 5, the equation accounted for about 99% of the variance; in general, the equation accounts for 90% to 93% of the variance in consumption in studies conducted to date (Foltin, in press; Hursh et al., 1988, 1989; DeGrandpre & Bickel, personal communication). This method of analysis may have applications for assessing the abuse liability (likelihood of human acquisition and maintenance of drug taking) of new drugs prior to use in humans. Drug self-administration studies with nonhuman primates are commonly used to establish the reinforcing properties of drugs and to provide a qualitative assessment of the strength of performance maintained by them (Brady & Lukas, 1984; Johanson, 1978). Demand curve analysis based on an application of Equation 1 under standardized testing conditions could provide a quantitative assessment of abuse liability. The parameters of the demand equation provide an estimate of the expected level of consumption (L) and the sensitivity of consumption to increases in price (a). If other aspects of the experiment are held constant (e.g., effort of the response, unit of consumption, and the length of daily test sessions), then abuse liability of different drugs could be compared based on the values of the parameters of the demand equation. Demand curve analysis is not time consuming. Once stable self-administration is established at a low FR (e.g., FR 2), a demand curve can be estimated by increasing the FR in 20% to 40% steps each test day up to a maximum FR that reduces consumption to about 10% of baseline, a procedure that takes about 20 to 30 days. The demand curves in Figures 4, 13, and 14 were obtained using this method. Several repetitions of this procedure may be used to establish the stability of the parameter estimates in time. Other studies have used test sequences with even larger step sizes each day that allow exploration of the range of consumption in as few as 7 test days (see Figure 8; see also Hursh et al., 1988; Raslear, Bauman, Hursh, Shurtleff, & Simmons, 1988). Given the speed of the assessment, it is feasible, in the same test subjects, to examine the modification of demand by competing reinforcers or substitution of other drugs and to quantify the effects in terms of changes in the parameters of Equation 1.

BEHA VIORAL ECONOMICS OF DRUG ABUSE Demand for Pentobarbital/ Ethanol Mixture 1000

Responding for Pentobarbital/ Ethanol Mixture 8,000

c 0

LU 0 cz x

z0 cLI

6,400

CLJ

vn

100

4,800

z 0

° 3,200

Av

0

w

0 2 Deliveris

0 I-

o

1,600

8 Deliveries w

A 16 Deiverie

.

I

10

100

1

UNIT PRICE

10 UNIT PRICE

32

I100

Fig. 5. Left panel: For a rhesus monkey, total daily consumption of pentobarbital/ethanol mixture as a function of unit price (see text for explanation), in log-log coordinates. Right panel: For same conditions in left panel, average daily responses per day as a function of unit price, in semilog coordinates. In each panel, Pma. is indicated by vertical grid lines at a unit price of 32. Data from Lemaire and Meisch (1985).

Calibration methods. The definition of unit price and the mathematics of Equation 1 suggest that any comparison of elasticity estimates across laboratories will require an estimate of the effort of each response and calibration of the units and value of each reinforcer. This presents a problem for any retrospective analysis of demand from previous studies with drug reinforcers; precise comparison of elasticities will be inaccurate without measurement of response effort and comparable units of dose and potency.

Future studies can correct for these methodological differences. The most direct approach to calibration of effort would be to adopt a standard, calibrated operandum for these studies. Standardization may not be necessary, however, if appropriate measurements are taken to permit computation of a correction factor. A recent study by Bauman (1991) indicated that response-initiated fixed-interval schedules reduced consumption about as much as fixed-ratio schedules that were executed in the same amount of time. By implication, the results support a proposal that the amount of time taken to complete an FR is the critical cost factor that decreases consumption. This finding, coupled with the observation by Hursh et al. (1989) that doubling effort had the effect of doubling the median interresponse time and the average time to complete a given FR, suggests that the time required per unit reinforcement may conveniently summarize the effects of both effort and FR. This easily obtained measure, then, could be used to calibrate cost across different laboratories for the compari-

of demand elasticity; however, this proposal requires further systematic investigation. Likewise, and perhaps more importantly, the size and value of the reinforcer must be calibrated across experiments. Because reinforcer size and value enter into the calculation of both unit price and daily consumption, differences in measurement can lead to erroneous differences in the parameter estimates of demand and elasticity. Correction for units of dose is an easy task; more difficult is calibration of reinforcer value, because drugs often differ in pharmacological potency. Hursh et al. (1989), in a study using food reinforcement, provided a method for estimating differences in reinforcer value from the observed levels of consumption at FR 1. In brief, the ratio of baseline levels of consumption was used to adjust unit price to obtain a value-independent measure of elasticity. For example, if Reinforcer A supported twice the consumption at FR 1 as Reinforcer B, then A is estimated to have half the potency of B (QB/Q.). This correction can be made empirically. In the comparison of demand for food and saccharin shown in Figure 4, the saccharin reinforcer was adjusted in volume to yield a daily reinforcer consumption equal to that of food at FR 1. Having equated the reinforcers for value (level of consumption at FR 1), demand could then be fairly compared in terms of reinforcers per day as a function of responses per reinforcer. In the hypothetical example given here, Reinforcer A could be dispensed in reinforcer units twice that of Reinforcer B to yield equal value units of reinforcement at FR 1. Elasticity son

STEVEN R. HURSH

382

0

I

-0

10

,,cT BE

*

LUJ 0-

0

C,,

courtwme

* FOOD O SUCROSE

ToBITS

Fig. 6. Diagram of four hypothetical forms of reinforcer interactions (see text for explanation).

1

1

UNIT PRICE of demand for A and B could then be directly compared without concern for reinforcer size or potency. Additional work with drug reinforcers is required to validate these methods. DRUG-REINFORCER INTERACTIONS Within a behavioral economic framework, reinforcer interactions are classified into several categories, illustrated in Figure 6. Most studies of choice with nonhumans have arranged for the alternative behavior to provide the same, perfectly substitutable reinforcer, usually food. This yields a specific kind of interaction in which the amount of behavior

100

10 -

FOOD

Fig. 8. Mean daily consumption, by 6 rats, of food and sucrose as a function of the unit price (FR schedule) for food, in log-log coordinates.

to each roughly matches the amount of reinforcement received from each (the matching law; see Davison & McCarthy, 1988). When the two alternatives require a specific number of responses per reinforcer delivery, subjects generally show exclusive preference for the least costly of the alternatives (Herrnstein, 1958; Herrnstein & Loveland, 1975). This situation is much like comparison shopping for identical items from different stores; all else being equal, one will go to the store with the lowest price. Most choices are between commodities that are not perfect substitutes. The other interm actions depicted in Figure 6 are imperfect substitutes, complements, and independent rein25 forcers. Figure 7 illustrates the difference 0 between imperfect substitutes and complements. Along the x axis is the price of Com0 modity A; along the y axis is the quantity of consumption of the alternative Commodity B with fixed price. As the price of A increases, z 0 consumption of A decreases, the usual demand relation. If, at the same time, the consumption a. of B increases in response to these increases in the price of A, then B is defined as a substitute for A. If the consumption of B decreases, then z Co Cj) 0 B is defined as a complement of A. 10 Choice between two imperfect substitutes is 10 100 illustrated in Figure 8 (Hursh & Bauman, UNIT PRICE - COMMODITY A 1986). One alternative was a nutritive food Fig. 7. Diagram of hypothetical changes in consump- pellet available after pressing a lever; the other of tion of Commodity B as a function of the unit price Commodity A. Circles indicate a complementary relation; alternative was pellets of sucrose freely available in a cup. The food pellet was a balanced squares indicate a substitutable relation. I

BEHA VIORAL ECONOMICS OF DRUG ABUSE 10001

A

AA

AA

AA

A

A

AAA

AA AA

COD

A A AA A

00 0

Cl:

383

0

0 0

0

et

A 0

0

COO C03 C= F-

g C013 100Co 0

0

00

Co -J

UJ A

WATER FR 10

0

00' IO FOOD 10

100

1000

FIXED RATIO

Fig. 10. Diagram of nonreciprocal interaction beethanol and phencyclidine (PCP). Features of ethanol may substitute for PCP, but features of PCP may not substitute for those of ethanol that are outside its domain of stimulation (shaded area).

Fig. 9. For a representative monkey, daily consumption of food and water as a function of the FR for food, in log-log coordinates. Water was constantly available under an FR 10 schedule.

tween

diet, but less sweet; the sucrose was sweet and caloric but lacked many essential nutrients. The price of the food pellets was increased across conditions of the experiment; daily consumption of sucrose was measured at each price of food. Even at the lowest price of food (FR 1), some sucrose was consumed; as the price of food increased and food consumption decreased, consumption of sucrose increased as a partial substitute. However, even at the highest price of food (FR 243), some food was consumed despite the free availability of sucrose. This reciprocal trade-off between consumption of two reinforcers is typical of imperfect substitutes. Choice between two complements is illustrated in Figure 9. One alternative provided pellets of food; the other alternative provided squirts of water. As the price of food increased, decreasing food consumption, daily consumption of water decreased. The value of water as a reinforcer declined as the consumption of food declined. Drug reinforcers may, by their neurochemical nature, reflect different forms of interaction that parallel these interactions seen with consumable reinforcers. An excellent example of different drug interactions was demonstrated by Carroll (1987). In this study subjects were given two drug alternatives, phencycli-

dine (PCP) or ethanol (ETOH). The price of each commodity was varied by altering the concentration of drug in each delivery; lower concentrations represented higher unit prices of the drug. In the first experiment, the unit price of PCP was varied, once without ETOH available and later with ETOH available. As the unit price of PCP increased, consumption of PCP declined, and the decline was greater when ETOH was available as a substitute. At the same time, ETOH consumption increased at the highest unit prices of PCP. Thus, ETOH functioned as a substitute for PCP as the price of PCP increased. In a second experiment, the unit price of ETOH was varied, once without PCP available and later with PCP available. As the unit price of ETOH increased, consumption of ETOH declined, but the rate of decline was independent of the availability of PCP. Furthermore, there was little change in the consumption of PCP as the consumption of ETOH was driven down by increasing price. Thus, although ETOH served as a substitute for PCP in Experiment 1, in Experiment 2 PCP did not serve as a substitute for ETOH. This study illustrates a nonreciprocal interaction between two commodities. Figure 10 depicts diagrammatically how such an asymmetry might come about. Conceive of each drug reinforcer as consisting of

384

STEVEN R. HURSH

a group of stimulus features, perhaps coinciding with areas of central nervous system stimulation. If we think of ethanol as a drug with a relatively large set of stimulus features (it has nonspecific action throughout the brain) and PCP as a drug that stimulates only a subset of those same features (it activates specific neural receptors), we can see that ETOH stimulation would substitute for PCP stimulation of the subset, but that PCP stimulation could not substitute for those stimulus features of ETOH that were outside the PCP domain of stimulation. Thus, observations of reciprocal and nonreciprocal reinforcer interactions may give some clue to the underlying neural mechanism of action of the two alternative drug reinforcers. DRUG USE INTERVENTIONS The concepts of substitution and complementarity provide some insights into important limitations of individual therapy programs for the control or elimination of drug abuse in individual clients. Within a behavioral framework, one can conceptualize the therapeutic situation as one in which the therapist or clinician attempts to shape new behavior under the control of acceptable reinforcers that compete with and reduce the occurrence of behavior to obtain illicit drugs. Thus, the reinforcers arranged by the therapeutic process interact with those from the use of illicit drugs (see Thompson, Koerner, & Grabowski, 1984). The effectiveness of this competition will depend, at least in part, on several economic factors: the amount of direct substitution between the two sources of reinforcement, the availability of desirable complements to the therapeutic reinforcers that will maximize their effectiveness, and the amount of direct competition that exists between the two sources of reinforcement (i.e., does performance for one preclude or prevent reinforcement from the other). These factors can be illustrated by considering the effectiveness of methadone therapy for users of heroin. Methadone is an imperfect substitute for heroin; it is usually explicitly formulated so that an oral dose will prevent opiate withdrawal but will not produce a pronounced euphoria or "high." It substitutes for heroin to prevent withdrawal symptoms, the aversive consequences of nonuse of the drug, but does

not substitute for the immediate positive reinforcing consequences of euphoria. One could predict, then, that even if a large price differential existed between the two commodities, some heroin would still be purchased from illicit sources for its unique reinforcing features (see Stitzer, Grabowski, & Henningfield, 1984). In addition, heroin is often consumed as part of a social ritual, and these social events serve as complements to the primary reinforcing consequences of the drug. To the extent that the substitute, methadone, must be consumed in a clinical, nonsocial environment, its value will be diminished as an adequate substitute for heroin because it is not accompanied by important complementary social reinforcers (see Hunt, Lipton, Goldsmith, & Strug, 1984). Therapy for opiate addiction usually consists of other reinforcers along with methadone, such as ajob and participation in a therapy group. The effectiveness of these alternative reinforcers as deterrents to further illicit drug use will depend, in part, on the existence of direct competition between the two sources of reinforcement. A study by Elsmore, Fletcher, Conrad, and Sodetz (1980) illustrates the importance of competition in reducing drug consumption by a nondrug reinforcer. In this experiment, baboons were intermittently presented with two keylights that signaled the availability of a choice between an intravenous infusion of heroin (0.1 mg/kg heroin HCI) or delivery of 3 g of food (four 750-mg Noyes pellets). Trials were separated by an interval that was varied across conditions from 2 min to 12 min. Subjects lived in there test cages and received all their food under the test conditions. The results are shown in Figure 11. When trials were plentiful (every 2 min), frequent choices of both food and heroin occurred; the mere presence of the alternative food reinforcer did not eliminate heroin consumption. However, as the frequency of trials decreased with longer intervals between choices, heroin choices were placed in competition with maintenance of food consumption. Under these conditions of reduced "income" of trials, the subjects allocated proportionately more of the available trials to food and less to heroin, as one might expect from the differences in elasticity shown in Figure 1. Under the conditions with most infrequent trials (every 12 min), this

BEHA VIORAL ECONOMICS OF DRUG ABUSE 100 - -

--.&- -------- --_k-0~~~~~~

50

-

°

--

-

-

~~~~~~Foo

11-

A 0 0

4

a s

IL

LA

0

zU

Food

20-

Baboon P241

a

Heroin A

Baboon P363

0

0

10 A

4

2

4

8

12

INTERTRIAL INTERVAL (Min.)

Fig. 11. For 2 baboons, choices per day for food and heroin as a function of the intertrial interval. Best fitting straight lines are drawn through the points. Data are from Elsmore et al. (1980).

competition reduced consumption of heroin an average of 83% compared to conditions with frequent trials. This finding of an income effect on choice behavior has been confirmed in several other studies with primates, mice, rats, and pigeons (Hastjarjo, Silberberg, & Hursh, 1990; Sakagami, Hursh, Christensen, & Silberberg, 1989; Shurtleff & Silberberg, 1990; Silberberg, Warren-Boulton, & Asano, 1987). These laboratory results with nonhuman subjects suggest that procedures that require heroin abstinence as a prerequisite for retention of a job or admittance to the therapy group might be expected to enhance the direct competition between the drug and nondrug reinforcers. Of course, unlike the previously described study with food pitted against heroin, the winner of that competition may not be therapeutic alternatives, given the other limitations of therapy listed above. However, given sufficiently attractive therapeutic alternatives, such competition may be critical to elimination of further illicit drug use. This may be especially true for beginning drug users prior to the development of tolerance and dependence as a consequence of repeated drug exposure, and prior to isolation from the community of

those who do not use drugs and integration into the social network of drug users (Hubbard, Rachal, Craddock, & Cavanaugh, 1984; see Grilly, 1989, for a review). DRUG ABUSE POLICY Specific therapeutic interventions may be viewed as one part of an overall public policy concerning illicit drug sale and use. Behavioral economics may be used as an important interpretive and analytic tool for evaluating alternative public policy initiatives. The first requirement of this strategy is the clear specification of public policy objectives. A partial list of those objectives might include the following: (a) reduction of illicit drug use, (b) reduction of crime by drug users, (c) reduction of distribution and sale of illicit drugs, and (d) reduction of distribution-related crime. These objectives may be restated in behavioral economic terms. In order to make this translation, first consider the total drug marketplace as represented by the demand and supply interaction diagrammed in Figure 12. Demand has been previously defined as the level of consumption that will occur across a series of prices; supply describes the behavior

STEVEN R. HURSH

386 MARKET EQUILIBRIUM

DEMAND

SUPPLY

MINIMIZE PRICE

1-11 /

MAXIMIZE PRICE

/

Fig. 12. Diagram of hypothetical market equilibrium between consumer demand and seller supply of a particular commodity, as a function of market price. Consumer attempts to minimize price; seller attempts to maximize price. The equilibrium market price is shown as the intersection of these two hypothetical functions.

of the seller of the drug and indicates that higher prices increase the likelihood that individuals will obtain and resell the illicit drug to the end user. According to microeconomic theories of price (Watson & Homan, 1977), these two forces are in competition in the marketplace, and the equilibrium market price is determined by the intersection of these two functions. The level of illicit drug use is simply the total level of drug consumption at the equilibrium market price. The level of illicit drug sale and distribution is that portion of total drug consumption that is from illicit sources. Crime related to drug use, for the most part, can be ascribed to two primary factors-the need for money to purchase the drugs, on the one hand, and the loss of self-control that is a direct result of drug use. The need for money may be quantified as the total daily allocation of performance (money) for the purchase of the drug, exemplified in Figure 5 as total responses per day. These same dollars, if exclusively spent on illicit drugs, represent the total revenues to the drug sellers. Distribution-related crime, such as smuggling and violent competition among dealers, likely reflects the size of this revenue pool that serves as the reinforcer for engaging in these activities. Public policy influences on this process may be categorized as those that focus on reducing supply (supply-side interventions) and those that focus on reducing demand (demand-side

interventions). These two aspects of the process are inseparable, and public policy should not be seen as a choice between them. Indeed, it will be shown below that maximum effects may require a coordinated application of both kinds of policies. Supply-side interventions. Supply-side interventions are commonly used by government agencies in the United States, and include criminal punishment for production, importation, distribution, and sale of illicit drugs; confiscation of stockpiles; interdiction of imports; destruction of crops (poppies, coca, and marijuana); and controls on chemical production plants that provide the raw materials for the drugs. All of these interventions are designed to increase the costs associated with the supply process, thus moving the supply function to the right. Given a constant level of demand, this well have the effect of increasing the equilibrium price. To evaluate the possible effects of this change in the market, consider the demand curve for pentobarbital/ethanol depicted in Figure 5, which is representative of demand curves for most drugs studied in the laboratory (Bickel et al., 1990). Demand is inelastic across a broad range of prices, here changing to elastic at a unit price of 32. All increases in price will tend to reduce consumption (Objective 1), but for price increases in the inelastic region, the decreases in consumption will be small relative to the changes in price. In the study illustrated in Figure 5 (Lemaire & Meisch, 1985), an eight-fold increase in unit price from 4 to 32 produced only about a 55% reduction in consumption. The response-rate function shown in Figure 5 represents the "total revenues" generated by the demand curve and indicates that across the range of unit prices up to 32, revenues to the sellers increase. Above that price, revenues decrease. Given this inverted U-shaped function for total responding, the effect of a price increase on total expenditures and revenues to the suppliers will depend on the initial market price, be it in the inelastic or elastic range of the demand curve. Only if the initial equilibrium price is already quite high (i.e., equal to or greater than P..) will such a change result in reduced expenditures and revenues. Otherwise, increased drug prices will increase total expenditures, which will, in turn, tend to generate more drug-related crime to obtain money to meet the higher prices. At the same time,

BEHA VIORAL ECONOMICS OF DRUG ABUSE

387

RESPONSE OUTPUT

DEMAND

0 0

1000

10000I > 100 0 cc w oL

0

0

ccLU

100

cn C,

COw

CO) z

-J -J

100

0

I

CO)

0

0 0 LL

NO FREE 0

cc i'U

PRICE ( FIXED RATIO) Fig. 13. Left panel: Median of 3 rhesus monkeys showing daily food earned in the work session as a function of price (FR schedule). Three curves are for conditions with no free, or with either one third or two thirds of the baseline level of consumption at FR 1 provided free after the 12-hr work session. Curves through the points are the best fit of Equation 1. Right panel: For same conditions in left panel, the total daily responses emitted for food in the work session, along with the best fitting curves derived from Equation 1. Both panels are in log-log coordinates.

increased supplier revenues will increase the incentive to engage in supplier-related activities. These outcomes are all counter to Objectives 2, 3, and 4 stated above. This kind of unintended consequence has been observed several times in the petroleum industry; when petroleum suppliers attempt to reduce production and force up prices, the inevitable consequence is an increase in the revenues (and profits) of the oil distribution companies, because demand for petroleum by western countries is relatively inelastic. Demand-side interventions. The limitations of price adjustments for influencing drug consumption when demand is inelastic suggests that a more effective policy would incorporate methods to alter demand (see Jarvik, 1990). The previous section on individual drug therapy described approaches to lowering the overall level of demand by the individual consumer. In this section, a public policy approach to increase the elasticity of demand will be described. This approach can be best understood by analogy to an experiment the author conducted to alter the elasticity of demand for food by providing a "free" source of food after the

work periods (Hursh et al., 1989). Monkeys lived in test chambers that provided their total daily ration of food and water. In the first phase of the experiment, each work day started at 6:00 a.m. and ended at 6:00 p.m.; during the work period pressing a push-plate provided food according to an FR schedule (water was constantly available). The price of food was increased across 21 conditions from an initial prices of 10 responses to a maximum price of 372 responses per food pellet. At the end of the 12-hr work period, free pellets could be delivered. Three conditions were studied: one free pellet (baseline), one third of a normal ration was free, and two thirds of a normal ration was free. Free pellets were delivered response independently every 3 s immediately after the end of the 12-hr work period (signaled by a change in key colors). Consumption across the 21 price levels yielded a demand curve under each of the three conditions of free food. The results are shown in Figure 13. The median consumption of earned pellets is shown in the left panels and displays a graduated increase in curvature or elasticity with increases in the amount of free food. The re-

STEVEN R. HURSH

388 DEMAND

RESPONSE OUTPUT

100,000

NO FREE

10,000

_

>-

a

O

0

a

LL

cn

CEO

CO

ILJ

LJ 11

O 001000

a

(j)

ONE FREE MEAL

a.

8

0

0

w

Cc

1000 0

FOUJR FREE MEALS

10

100

1000

PRICE (FIXED RATIO) Fig. 14. Median of 3 rhesus monkeys showing food earned in four daily 1-hr work sessions as a function of the price (FR schedule). Curves are from conditions with no free food, one free meal, or four free meals (see text for explanation).

sponse-rate functions in the right panels show the usual inverted U-shaped pattern, with the peak of the curve shifting to the left (lower prices) with increases in the amount of free food. In a second phase of the experiment, the work day was divided into four work periods, each 1 hr long, distributed from 6:00 a.m. to 5:00 p.m. In the baseline condition, no free food was presented; in two free-food conditions, either a 20-min free meal (a period of food available at a price of one response per pellet) was presented at the end of the 12work day or four 5-min free meals were presented after each 1-hr work period. This last condition was intended to further increase substitution by reducing the maximum delay to food. The results of this phase are shown in Figure 14. The demand curves for earned pellets show a further increase in elasticity with the four free meals after each 1-hr work period. The lines through the points in Figures 13 and 14 were fitted using Equation 1 and accounted for an average of 93% of the variance. As expected, parameter a showed a consistent increase from 0.004 with no free food to 0.011 with four free meals, indicative of increased

curvature of the demand function. Figure 15 is a composite of the fitted curves showing the orderly increase in elasticity with increasing amount and immediacy of free food and the orderly shift to lower prices of the peak of the response output functions (Pm., indicated by

arrows). Providing free food in this experiment is analogous to providing the community of drug users a low-cost or free source of drugs, such as through government-sponsored clinics. The effects of such "medicalization" of drug use may be inferred from these results with free food. Inspection of the functions in Figure 15 indicates that at prices above about 50 responses per pellet, free food had the effect of reducing earned consumption at any given price compared to baseline and reducing total revenues. Further inspection of the response output functions indicate that even if the market price is adjusted to the peak of the response output (revenue) function (indicated by arrows), total revenues will drop dramatically (here nearly a factor of 10) with increases in free availability of drugs, indicated by horizontal grid lines. This reduction in expenditures and revenues would be expected to reduce drug-related crime (Objective 2), reduce

BEHAVIORAL ECONOMICS OF DRUG ABUSE DEMAND

389

RESPONSE OUTPUT

COMPOSITE

COMPOSrTE

1000

, 100

a

a

L- 1 0,000

cc

w

Ca

LUJ

CO

z

-10 w -J

0 X

.

CD) cr

-J w 0-

-,

100

1000

FIXED RATIO

1000

-

10

100

1000

FIXED RATIO

Fig. 15. A composite of the fitted functions from Figures 13 and 14 showing daily earned food (left panel) and total daily responses in the work session(s) (right panel). Curves are, from top to bottom, no free (12-hr session), one third free, two thirds free, one free meal, and four free meals. In the right panel, Pm.x is indicated by arrows, and the peak level of responding is indicated by horizontal grid lines.

the incentive to supply drugs (Objective 3), and reduce the incentive to engage in risky distribution-related crime (Objective 4). Notice, however, that all these benefits presume that the market price is relatively high and beyond the point of divergence of the demand curves, here a price of 50. This requirement argues for continued use of supply-side law enforcement measures in combination with this demand-side intervention to keep the illicit price high enough that illicit demand will be sensitive to the availability of the free source. Unfortunately, these positive outcomes come at some cost. Consider in the free-food experiment just described the total level of food consumption from both sources of food, earned plus free, shown in Figure 16 (data from a representative monkey). In all but one case for this subject, total food consumption was higher with free food available than without; in general, the subjects combined earned food and free food to "overeat" across all prices of earned food. In fact, these subjects generally became quite fat, judging from their body weights. By analogy, one might expect that in a condition of both illicit drug availability and clinically dispensed free drugs, overall drug use will increase, at odds with Objective 1. This would have the adverse effects of potentially expanding the base of drug users, increasing the general level of their use, and further reducing the productivity of society. In addition, those crimes that are a direct result of drug-distorted

behavior would also be expected to increase, such as auto accidents, suicide, and assaults. Thus, this approach to reducing illicit drug demand would best be combined with an aggressive and effective program of individual therapy to reduce the overall level of drug demand (see above). BEHAVIORAL UNKNOWNS This explicit and systematic analysis of the factors contributing to the demand for illicit drugs serves to highlight some significant uncertainties for public policy planning that need further investigation. Demand elasticity. As indicated above in the discussion of the effects of price increases on drug consumption and total expenditure, it is necessary to know the elasticity of demand in the vicinity of the current market price to make a precise prediction of the sensitivity of consumption to price increases and the direction of change in total expenditure of performance (money) to obtain the drug. This is because demand elasticity is not constant, and total expenditures (and supplier revenues) follow an inverted U-shaped function. If the market price is optimally positioned at the point of maximum revenue for the supplier (P..), then price increases will lead to proportional decreases in consumption and reductions in supplier revenues. On the other hand, if the market price is well within the inelastic range, then price increases will have a proportionately

STEVEN R. HURSH

390 FREE FOOD

BASELINE

a

cc L a. cD LLI -J

CL 0

0 0 LL

100

10

100

PRICE IN WORK SESSION Fig. 16. For a representative rhesus monkey from the studies illustrated in Figures 13 and 14, the total daily consumption of food from both work sessions and free sources, in log-log coordinates. Baseline consumption, shown as the heavier line, is without any free food; open symbols indicate total consumption including free sources.

smaller effect on consumption and will lead to in user expenditures and supplier

an increase revenues.

Similarly, the effectiveness of a governmentsponsored medicalized program of distribution to increase demand elasticity will also depend on the market price (see Figure 15); if market price is low, relative to the new alternative, one might predict little, if any, change in demand elasticity and the frequency of use of illicit drugs. Thus, without knowledge of consumer demand and market price, there is no way to predict with any confidence whether specific public policy interventions will increase or decrease drug use, drug-related crime, and supplier revenues. The formulation of public policy in the absence of this information is analogous to formulating tax policy in the absence of knowledge of the tax base and seems at best ineffectual, and at worst, counterproductive. Unfortunately, the inverted U-shaped expenditure-revenue function virtually guarantees that drug policy will be controversial. For any selected level of drug expenditures, there are two prices that produce the same level of total funds expended, a low price and a very high price. The choice between policies, then,

may hinge on judgments about other factors that accompany movement toward lower or higher prices, such as the overall level of consumption and related effects on society, the cost (both financial and social) of government interventions, and ethical issues, such as those associated with a program of medicalization that would use tax revenues to subsidize habits that many regard as offensive, immoral, or unhealthy (see Gallup & Gallup, 1988). An economic analysis makes a positive contribution by disentangling these issues from potentially erroneous presumptions about the economic processes that govern the drug market. Supply elasticity. The effectiveness of government-sponsored competition would be modulated by the elasticity of supply and the capacity for market prices to adjust to changes in market conditions. If wholesale costs are relatively low, if the threat of punishment is low, and if the supply of illicit drugs or raw material is plentiful, then the market price of illicit drugs may be able to adjust downward to maintain a competitive edge. Furthermore, market prices do not have to match those offered under a government-sponsored program because of the inherent nonsubstitutability of the two sources. The government clinic would provide the benefits of a pure drug at a reliable dose, but the illicit suppliers can make their product more readily available at any hour and without arbitrary limits on dose. Judging from the effects of time delay shown in Figure 14 (one vs. four free meals), an illicit source that provides more immediate availability in a socially supportive environment would offer stiff competition for a government clinic that may have set hours of operation, certain limits on daily dosage, and other bureaucratic inconveniences (forms to be filled out, painful blood tests, and waiting lines). Product innovation. Further, the drug supply system should be viewed as an active competitor; it is unlikely that illicit suppliers will stand idle while they are put out of business by government-sponsored clinics. In the legal marketplace, when competition from a new source occurs, one typical response to prevent effective competition is to engage in product innovation, that is, make your product distinct and nonsubstitutable from that available from the competition. Advertising focuses on these unique features and persuades the consumer that the competition provides an inferior prod-

BEHAVIORAL ECONOMICS OF DRUG ABUSE uct. In the drug trade, this kind of innovation will lead to new "designer drugs" or drug combinations with higher abuse potential, a stronger "kick," or longer lasting effects, with little concern for long-term user safety. Aggregate versus individual demand. During the foregoing discussion, we have extrapolated predictions about market demand from observations of individual demand. This assumes that the aggregation of demand curves from individuals introduces no new processes or variables. There are two ways in which this assumption may not be valid. First, the elasticity of demand of an aggregation of demand curves may not be as inelastic as each individual curve; this is an empirical question. Second, the level of demand of the aggregate is not fixed, as is the level for any given individual. Thus, one response to government policies to reduce individual demand is to expand the market to new communities where resistance to drug use is lower and availability of government competition is low. These new markets will, in turn, force further government involvement and an escalation in the cost of the government-sponsored program. Supplier competition. Reductions of demand and pressures to increase price will likely lead to heightened competition among drug suppliers for the diminishing market. In the legal marketplace, such competition is expressed in sanctioned marketing techniques-advertising, sale prices, rebates, and product innovations. In the illegal marketplace, less benign forms of competition prevail, including murder, extortion, and product fraud. All these byproducts must be anticipated by any effective public policy intervention that reduces demand and/or increases price. A STRATEGY FOR PUBLIC POLICY FORMULATION It is clear that an empirical approach to

public policy formulation in the sphere of drug abuse will require an extensive knowledge of both the principles of behavior and the prevailing dynamics of the illicit drug market. Figure 17 diagrams an overall strategy for public policy formulation and implementation. At the primary level is the need for expanded basic research on the principles that govern drug use from laboratory and clinical settings. Based on the knowledge of these basic processes, we can then formulate the critical un-

Laboratory Non-Human

Laboratory H n

391

Clinical Research

Fig. 17. Flow chart illustration of behavioral economic strategy for public policy formulation and implementation based on data and analysis from laboratory, clinical, and econometric studies and from practical applications in experimental model projects.

knowns that are important for prediction of policy outcomes. This establishes the goals of an econometric analysis of actual drug use in the natural marketplace. A program of monitoring of drug-use patterns and market prices would establish approximations of demand elasticity for prediction of policy impacts. However, even the most ambitious econometric program will be difficult and prone to error because the details of supply, distribution, and use are deliberately hidden to avoid criminal prosecution. Thus, any public policy formulation based on this knowledge must be considered tentative. To avoid error on a large scale, experimental model projects should be undertaken to test creative solutions and novel public policy initiatives. Coupled with each such project should be an extensive program of evaluation and improvement. Only after several iterations of evaluation and modification should the methods of a model project be proposed for general application. Even then, because of the uncertainties of scale and the diversity of settings, public policy must be formulated and implemented with provisions for self-evaluation and correction without the need for new legislation. The strategy envisioned is a self-correcting one. The results of evaluations serve as feedback to the policy formulation process and to the research community for needed investigations to improve approaches. Provisions are

STEVEN R. HURSH

392

needed to insure that the system can adjust rapidly to this feedback. Studies of behavior in the laboratory clearly show that delays imposed between changes in behavior and changes in the controlling environment can lead to unintended effects on subsequent behavior. Therefore, an effective behavioral economic system for influencing the dynamic drug market will require an especially responsive bureaucracy with minimal feedback delays. CONCLUSIONS Behavioral economics may be defined as a special application of behavior analysis that emphasizes environmental and biological factors modulating the total allocation of performance to available reinforcers. When the commodities of interest are drug reinforcers, this method of analysis provides a consistent conceptual framework for quantifying the strength of drug self-administration in laboratory subjects, for evaluating interactions between drug and nondrug reinforcers, for designing and interpreting the effectiveness of drug abuse interventions in humans, and for formulating public policy to deal with drug abuse. These separate domains can be seen as part of an overall strategy that would draw on the findings from laboratory research and naturalistic econometric investigations to design model intervention projects. Evaluation and improvement of these projects would then serve as small-scale test-beds for more general public policy initiatives. This empirical and behavioral approach to public policy in the sphere of drug abuse is representative of a more general behavioral system for the design of culture, as anticipated by Skinner (1953). If future work confirms the utility of this behavioral economic agenda, then we have made the first step toward the development of a more humane society.

REFERENCES Bauman, R. A. (1991). An experimental analysis of the cost of food in a closed economy. Journal of the Experimental Analysis of Behavvior, 56, 33-50. Bickel, W. K., DeGrandpre, R. J., Higgins, S. T., & Hughes, J. R. (1990). Behavioral economics of drug self-administration. I. Functional equivalence of response requirement and drug dose. Life Sciences, 47, 1501-1510. Bickel, W. K., DeGrandpre, R. J., Hughes, J. R., & Higgins, S. T. (1991). Behavioral economics of drug

self-administration. II. A unit-price analysis of cigarette smoking. Journal of the Experimental Analysis of Behavior, 55, 145-154. Brady, J. V., & Lukas, S. E. (Eds.). (1984). Testing drugs for physical dependence potential and abuse liability (NIDA Research Monograph 52). Rockville, MD: Department of Health and Human Services, National Institute of Drug Abuse. Carroll, M. E. (1987). Self-administration of orallydelivered phencyclidine and ethanol under concurrent fixed-ratio schedules in rhesus monkeys. Psychopharmacology, 93, 1-7. Davison, M., & McCarthy, D. (1988). The matching law: A research review. Hilldale, NJ: Erlbaum. Elsmore, T. F., Fletcher, D. V., Conrad, D. G., & Sodetz, F. J. (1980). Reduction of heroin intake in baboons by an economic constraint. Pharmacology Biochemistry and Behavior, 13, 729-731. Foltin, R. W. (in press). An economic analysis of "demand" for food in baboons. Journal of the Experimental Analysis of Behavior. Gallup, G., Jr., & Gallup, A. (1988). The Gallup Poll: Public opinion 1988 (pp. 124-128). Wilmington, DE: Scholarly Resources. Griffiths, R. R., Bigelow, G. E., & Henningield, J. E. (1980). Similarities in animal and human drug-taking behavior. In N. K. Mello (Ed.), Advances in substance abuse (Vol. 1, pp. 1-90). Greenwich, CT: JAI Press. Grilly, D. M. (1989). Drugs and human behavior. Boston: Allyn and Bacon. Hastjarjo, T., Silberberg, A., & Hursh, S. R. (1990). Quinine pellets as an inferior good and a Giffen good in rats. Journal of the Experimental Analysis of Behavior, 53, 263-271. Herrnstein, R. J. (1958). Some factors influencing behavior in a two-response situation. Transactions of the New York Academy of Sciences, 21, 35-45. Herrnstein, R. J., & Loveland, D. H. (1975). Maximizing and matching on concurrent ratio schedules. Journal of the Experimental Analysis of Behavior, 24, 107-116. Hubbard, R. L., Rachal, J. V., Craddock, S. G., & Cavanaugh, E. R. (1984). Treatment outcome prospective study (TOPS): Client characteristics and behaviors before, during, and after treatment. In F. M. Tims & J. P. Ludford (Eds.), Drug abuse treatment evaluation: Strategies, progress, and prospects (pp. 42-68). (NIDA Research Monograph 51, DHHS Publication No. ADM 84-1329). Washington, DC: U.S. Government Printing Office. Hunt, D. E., Lipton, D. S., Goldsmith, D. S., & Strug, D. L. (1984). Problems in methadone treatment: The influence of reference groups. In J. Grabowski, M. L. Stitzer, & J. E. Henningfield (Eds.), Behavioral intervention techniques in drug abuse treatment (pp. 8-22). (NIDA Research Monograph 46, DHHS Publication No. ADM 84-1282). Washington, DC: U.S. Government Printing Office. Hursh, S. R. (1980). Economic concepts for the analysis of behavior. Journal of the Experimental Analysis of Behavior, 34, 219-238. Hursh, S. R. (1984). Behavioral economics. Journal of the Experimental Analysis of Behavior, 42, 435-452. Hursh, S. R., & Bauman, R. A. (1986). The physiology and psychology of substitution in a closed economy.

BEHA VIORAL ECONOMICS OF DRUG ABUSE Proceedings and Abstracts of the Annual Meeting of the Eastern Psychological Association, 57, 49. (Abstract) Hursh, S. R., & Bauman, R. A. (1987). The behavioral analysis of demand. In L. Green & J. H. Kagel (Eds.), Advances in behavioral economics (Vol. 1, pp. 117-165). Norwood, NJ: Ablex. Hursh, S. R., Raslear, T. G., Bauman, R., & Black, H. (1989). The quantitative analysis of economic behavior with laboratory animals. In K. G. Grunert & F. Olander (Eds.), Understanding economic behaviour (Theory and Decision Library, Series A, Vol. 2, pp. 393-407). Dordrecht, Netherlands: Kluwer. Hursh, S. R., Raslear, T. G., Shurtleff, D., Bauman, R., & Simmons, L. (1988). A cost-benefit analysis of demand for food. Journal of the Experimental Analysis of Behavior, 50, 419-440. Jarvik, M. E. (1990). The drug dilemma: Manipulating the demand. Science, 250, 387-392. Johanson, C. E. (1978). Drugs as reinforcers. In D. E. Blackman & D. J. Sanger (Eds.), Contemporary research in behavioral pharmacology (pp. 325-390). New York: Plenum Press. Lea, S. E. G. (1978). The psychology and economics of demand. Psychological Bulletin, 85, 441-466. Lea, S. E. G., & Roper, T. J. (1977). Demand for food on fixed-ratio schedules as a function of the quality of concurrently available reinforcement. Journal of the Experimental Analysis of Behavior, 27, 371-380. Lemaire, G. A., & Meisch, R. A. (1985). Oral drug self-administration in rhesus monkeys: Interactions between drug amount and fixed-ratio size. Journal of the Experimental Analysis of Behavior, 44, 377-389. Rashotte, M. E., & Henderson, D. (1988). Coping with rising food costs in a closed economy: Feeding behavior and nocturnal hypothermia in pigeons. Journal of the Experimental Analysis of Behavior, 50, 441-456. Raslear, T. G., Bauman, R. A., Hursh, S. R., Shurtleff, D., & Simmons, L. (1988). Rapid demand curves

393

for behavioral economics. Animal Learning & Behavior, 16, 330-339. Sakagami, T., Hursh, S. R., Christensen, J., & Silberberg, A. (1989). Income maximizing in concurrent interval-ratio schedules. Journal of the Experimental Analysis of Behavior, 52, 41-46. Shurtleff, D., & Silberberg, A. (1990). Income maximizing on concurrent ratio-interval schedules of reinforcement. Journal of the Experimental Analysis of Behavior, 53, 273-284. Silberberg, A., Warren-Boulton, F. R., & Asano, T. (1987). Inferior-good and Giffin-good effects in monkey choice behavior. Journal of Experimental Psychology: Animal Behavior Processes, 13, 292-301. Skinner, B. F. (1953). Science and human behavior. New York: Macmillan. Stitzer, M. L., Grabowski, J., & Henningfield, J. E. (1984). Behavioral intervention techniques in drug abuse treatment: Summary of discussion. In J. Grabowski, M. L. Stitzer, & J. E. Henningfield (Eds.), Behavioral intervention techniques in drug abuse treatment (pp. 147-156) (NIDA Research Monograph 46, DHHS Publication No. ADM 84-1282). Washington, DC: U.S. Government Printing Office. Thompson, T., Koerner, J., & Grabowski, J. (1984). Brokerage model rehabilitation system for opiate dependence: A behavioral analysis. In J. Grabowski, M. L. Stitzer, & J. E. Henningfield (Eds.), Behavioral intervention techniques in drug abuse treatment (pp. 131 146) (NIDA Research Monograph 46, DHHS Publication No. ADM 84-1282). Washington, DC: U.S. Government Printing Office. Watson, D. S., & Holman, M. A. (1977). Price theory and its uses (4th ed.). Boston: Houghton Mifflin. Received May 18, 1991 Final acceptance May 29, 1991

Behavioral economics of drug self-administration and drug abuse policy.

The concepts of behavioral economics have proven useful for understanding the environmental control of overall levels of responding for a variety of c...
3MB Sizes 0 Downloads 0 Views