An Addiction Science Network Resource

Reprinted from J.R. Weeks and R.J Collins (1987), Screening for drug reinforcement using intravenous self-administration in the rat. In M.A. Bozarth (Ed.), Methods of assessing the reinforcing properties of abused drugs (pp. 35-43). New York: Springer-Verlag.
Brain Reward System
Back to book TOC
(Search book contents) (Search entire ASNet)

Chapter 2

Screening for Drug Reinforcement Using
Intravenous Self-Administration in the Rat

James R. Weeks and R. James Collins

The Upjohn Company
Kalamazoo, Michigan 49001

The reinforcing activity of 31 psychoactive drugs was evaluated. Drugs were offered by intravenous self-administration to groups of naive rats. Reinforcement is indicated by an injection rate greater than that for rats offered only saline. Two similar protocols were used. In the first protocol, rats were offered drug at a selected dose for 5 days, then the dose was reduced by one log unit (to 0.1 times the initial dose) for an additional 4 days. In the second protocol, saline only was offered the first 3 days to eliminate rats with high or low operant rates, and the dose was reduced 0.5 log unit (to 0.32 times the initial dose) instead of one log unit. An empirical score, based on the injection rates during the last 3 days of each period, describes the reinforcing activity. Replicate tests using both protocols gave similar scores. Literature data for reinforcing activity in monkeys was available for 27 of the drugs tested. Results for rats and monkeys agreed for 24 of the 27 drugs. Nalorphine and ethylketazocine were reinforcers only in rats, and ethanol was a reinforcer only in monkeys.



Assessment of abuse liability is an important consideration in the development of new psychoactive drugs. It is now well accepted that drugs which are reinforcing in animals by intravenous self-administration may have abuse liability in humans (Griffiths & Balster, 1979; Griffiths, Bigelow, & Henningfield, 1980; Griffiths, Brady, & Bradford, 1979; van Ree, 1979; Schuster & Thompson, 1969; Thompson & Unna, 1977). Monkeys have been the preferred species, and regulatory agencies in some countries specify testing in monkeys. Cost and limited availability of monkeys or other sub-human primates preclude testing of compounds early in the drug development process. A reliable method using rodents would be very desirable.

Any protocol for screening compounds for a pharmacological activity may be viewed as a special form of a biological assay. Both screens and bioassays require a dose-related assessment of the activity with adequate statistical controls. In a screen less precise measures are tolerable, and significance levels can be less stringent than the conventional 5% level. False positives are more likely to occur, but results from screens are followed with more testing to eliminate them. Implied in a screening test is the ability to test relatively large numbers of compounds without excessive cost or labor.

Several pharmacological classes of drugs are reinforcing by self-administration. When substances are tested early in the development process, knowledge of their pharmacology may be very limited. Accordingly, the screening method should be applicable without change across pharmacological classes. Drug-naive animals should be used.

We have recently devised a method for assessing reinforcing activity in rats (Collins, Weeks, Cooper, Good, & Russell, 1984). The method is suitable for several pharmacological classes of abused drugs, has a firm statistical basis, and gives a dose-related measure of reinforcing activity. Naive rats were offered test drugs by intravenous self-administration, without shaping or accessory cues, for several days and then the dose was reduced. Reinforcing activity was indicated by injection rates greater than those of control rats offered only saline. Activity was assessed in four ways: (1) an empirical score based on the injection rates for the two doses, (2) the number of rats reinforced (at either dose), (3) a comparison of the group mean injection rates to the rate for control rats self-injecting only saline, and (4) the presence of physical dependence. The data were gathered over a period of several years. At first we used a motor-driven syringe (Weeks & Collins, 1976). Syringes were often emptied when injection rates exceeded a few hundred a day. Later, a pneumatic-operated syringe with a fluid reservoir was developed which allowed unlimited numbers of injections (Weeks & Collins, 1981). After initial studies were completed, the protocol was modified to decrease the variability of the assay. These protocols will be referred to as "old" and "new," respectively. Thirteen drugs were re-tested using the new protocol. Since the protocols were very similar and the data were evaluated in the same manner, comparison of such results served to test the reproducibility of the method.

If the rat is a satisfactory alternate species for predicting abuse liability, results obtained must parallel those obtained in the monkey. Thirty-one different psychoactive drugs were studied. The reinforcing activity in the monkey had been reported in the literature for 27 of these drugs. Results in the rat were compared to these reports. The reader is referred to the publication by Collins et al. (1984) for details on interpretation of results and statistical validation of some aspects of the tests.

Materials and Methods


Rats were Sprague-Dawley origin females, ranging from 250 to 400 g. At least 6 days before use, rats were prepared with chronic venous cannulas (Weeks, 1972). After completion of the protocol, cannulas were considered functional if either blood could be withdrawn or the rat lost consciousness immediately following rapid injection of 3 to 5 mg/kg of sodium methohexital.


Experimental cages were standard individual hanging cages in a stainless steel rack similar to that used for prolonged intravenous infusions (Weeks, 1979). A lever switch (No. 121-03, BRS/LVE, 5301 Holland Drive, Beltsville, MD 20705), with the return spring removed to reduce operating pressure, was amounted on the rear wall of the cage 4 cm above the floor. It was essential to shield the paddle on the lever. Rats wore a saddle connected to a cannula feed-through swivel (see Weeks, 1977, for details of the lever switch and saddle). Each lever press resulted in a drug injection delivered by either a motor-driven or pneumatic syringe (Weeks, 1977, 1981; Weeks & Collins, 1976). The motor-driven syringe delivered injection volumes of 320 (initial dose) or 100 (reduced dose) ml/kg at 10.8 ml/second. The pneumatic syringe delivered only 100 ml/kg in about 0.5 second. Since 1979 we have used the pneumatic syringe exclusively (Ledger Technical Services, 2626 Lomond Drive, Kalamazoo, MI 49008). The total daily number of injections was noted at 8:00 a.m. 20 minutes, and any changes in experimental conditions were completed by no later than 9:30 a.m.

Most drugs were prepared in physiological saline. Drug forms and special solvents are noted in Collins et al. (1984). Doses of drugs were those of their respective salts, except for cyclazocine, ketamine, and pentazocine which were expressed as the free base.


The duration of the test was 13 days for both protocols. Rats were weighed initially, on return to the home cage after completion of the protocol, and again 24 hours later (initial, final, and withdrawal weights, respectively).

In the old protocol naive rats had been offered drug at an initial dose for 5 days, then the dose was reduced one log unit (to 0.1 the initial dose) for 4 days. A fixed-ratio (FR) schedule of FR-2 and FR-4 for two additional days each completed the protocol. Results for FR testing did not contribute to evaluation as a reinforcing agent and will not be discussed.

In the new protocol naive rats were offered 100 ml/kg of saline for 3 days, then the drug at an initial dose for 5 days, and finally the dose was reduced by 0.5 log unit (to 0.32 the initial dose) for 5 additional days. Rats averaging less than 4 or more than 50 injections per day of saline were rejected.

The dose selected for the first test was arbitrary, based upon known pharmacological effects of the drugs in rats. In subsequent tests, doses were increased or decreased in 0.5 log increments.

Controls were groups of 30 rats offered saline for 13 days. Separate controls were used for the motor-driven syringe old protocol, pneumatic syringe old protocol, and pneumatic syringe new protocol.

Evaluation of Reinforcing Activity

We showed earlier (Weeks & Collins, 1979) that daily injection rates of saline and morphine are log-normally distributed between rats. Analysis of 125 rats receiving only saline for 13 days supported the hypothesis that injection rates within rats are also log-normally distributed (Collins et al., 1984). Accordingly, each daily injection rate x was analyzed after the transformation log (x + 1) and the means expressed as the antilog - 1.

An empirical score was used to express the reinforcing activity of compounds in each individual rat. It was based on the injection rate for the last 3 days for the two drug doses compared to the same time periods for the appropriate saline controls. The 90% confidence limits for the number of injections of saline were calculated for both doses by the one-sample t-test.

Injection rates for rates self-administering saline and morphine
Figure 1: Injection rates and scores for rats self-administering saline and morphine using the pneumatic syringe and new protocol. The ellipse is the 90% confidence limits for saline control rats based upon the bivariate distribution of injection rates corresponding to the initial and reduced dose periods. The dashed lines represent the 90% confidence limits for saline self-administration for the initial and reduced doses individually. The scores for points falling in each quadrant formed by these lines are shown with the saline data. Open circles, score 0; solid triangles, score 1; solid squares, score 2; and solid circles, score 3. Note that injection rates are plotted to a logarithmic scale.

In addition, the 90% confidence ellipse, based upon the bivariate normal distribution, was calculated (Morrison, 1967). Rats whose mean injection rates were within either of these limits scored 0. A score of 1 signified that reinforcing activity was present only at the initial dose. A score of 2 signified that the reinforcing activity was present at both doses, therefore the initial dose was higher on the dose-response curve. A score of 3 signified that the initial dose was so high on the dose-response curve that responding was maintained at a rate not significantly greater than saline but became significantly greater when the dose was reduced. We have shown that self-injection rates of large doses of morphine fall within the saline range for many rats (Weeks & Collins, 1979).

Figure 1 illustrates the scoring system for saline and morphine doses of 0.01, 0.1 and 1 mg/kg on the new protocol. The dashed lines are the 90% confidence limits for saline injections for the initial and reduced doses, and the ellipse is the 90% confidence limit based upon the bivariate distribution. Three of the 30 saline rats fell outside these limits and so received a score, which would be expected at the 90% significance level. At 0.01 mg/kg of morphine, only one of six rats received a score. At 0.1 mg/kg, four of six rats were reinforced, each with a score of 2. The average daily injection rates for three of these four rats was over 250 per day for both doses, but the two non-reinforced rats took less than 30 injections per day. At 1 mg/kg all six rats were reinforced, four of them scoring 3. At the reduced dose all six rats took over 100 injections per day, but those rats scoring 3 took less than 25 injections per day on the initial dose.

Evaluation of Group Mean Injection Rates

If injection rates of most rats tested are only somewhat greater than saline, few if any would receive a score, yet the group mean could still be significantly greater than saline. The significance of this difference was calculated using Hotelling’s T2 statistic, which takes into consideration the bivariate analysis using data from both doses (Morrison, 1967).

Evaluation of Physical Dependence

A withdrawal weight loss significantly greater than that of saline controls was considered evidence of physical dependence. Significance was calculated using a one-tailed t-test.


Thirty-one psychoactive drugs were tested, 21 of them at more than one initial dose. Drugs were selected from several pharmacological classes. To illustrate dose-response relationships and correlation of results from the two protocols, drugs are divided into pharmacological classes (Johanson & Balster, 1978) and their mean scores plotted graphically (see Figure 2). Sixteen active drugs were tested at more than one dose. For some drugs further increases in the dose did not give a greater score, since the maximal effect was being approached. The relation to dose was not clear for methohexital, perhaps because lower doses had not been tested. Higher doses of methohexital might not yield valid results, since even at the doses tested some rats were nearly anesthetized. With those drugs which caused convulsions at toxic doses (e.g., meperidine, cocaine, methamphetamine, and nicotine), deaths from overdosage occurred at the higher doses tested.

The score versus log dose was analyzed as a linear regression for nine studies in which three or more doses of each drug were used. The slopes all had a common value (p = 0.98), and an estimate of the common slope (± SD) was 0.98 ± 0.1 (Collins et al., 1984). In other words, for each ten-fold (log 1) increase in dose the score increased about 1 unit.

Reinforcement scores for intravenous self-administration of various drugs
Figure 2: Reinforcing scores. Column 1: controls, narcotic agonists, mixed agonist-antagonists, and antagonists. Column 2: psychomotor stimulants, CNS depressants, and miscellaneous drugs. Saline doses are ml/kg. Protocol and syringe codes: solid = new, pneumatic; cross hatch = old, motor; diagonal = old, pneumatic. Reproduced with permission from Collins et al., 1984. Copyright 1984 by Springer-Verlag.

For mean scores greater than 1.3, mean injection rates for all rats taken as a group differed from saline (p = 0.01). However, scores are based on performance of individual rats, wherewith the injection rate must be three or more times greater than saline to achieve statistical significance. If several rats had injection rates greater than saline but not significantly different as individuals, the compound still could have reinforcing activity. Grouped data may resolve this uncertainty. In Figure 1 the probability for the mean injection rates differing from saline was 0.5 for morphine at 0.01 mg/kg. The score of 0.2 was probably due only to chance. However, for nalorphine (3.2 mg/kg, old protocol, motor syringe) the mean score was only 0.6, but the group data had differed significantly from saline (p = 0.003). Likewise, for ketamine (1 mg/kg, old protocol, pneumatic syringe) the mean score was 0.6, but the group data had been significantly different from saline (p = 0.001).

Results of a screening test must give consistent results when repeated. Generally two to three years elapsed between tests on old and new protocols. The two protocols differed only in minor details and the scores were calculated in the same manner. Analysis of the old and new protocol scores for the same drugs and doses as a linear regression yielded a correlation coefficient of 0.7, and neither the slope nor intercept differed significantly from unity (p > 0.05). The test method clearly yielded consistent, reproducible results.

Experimental data have been summarized in tabular form by Collins et al. (1984). For each drug, dose, and protocol, values are given for the mean injection rates, the mean weight changes, the number of rats reinforced out of the total tested, the significance level of the grouped data, and the mean scores. On specific request to R. J. Collins, we can supply the raw data.


Johanson and Balster (1978) have summarized world-wide tests for reinforcing activity of psychoactive drugs in monkeys using substitution procedures. Monkeys were trained to self-administer a reinforcing drug, such as cocaine or codeine, usually in 2- to 4-hour sessions. The test drug or vehicle was substituted, and the drug was considered active as a reinforcer when the injection rate for the test drug had been maintained at a rate significantly greater than when vehicle was substituted. When laboratories disagreed, we considered the drug active. In addition to the drugs in this report, ethylketazocine (ethyl cycloketazocine) was reported inactive (Woods, Smith, Medzihradsky, & Swain, 1979) and nicotine, although inactive in the substitution test described above, was active when offered by continuous self-administration (Yanagita, 1977). We considered nicotine as active in the monkey. In our rat studies, a drug was considered active if it met all of the following criteria at any dose: (1) at least half of the rats scored 1 or more, (2) the mean score was 1 or more, and (3) the significance level of the grouped data was 0.05 or less.

Results for monkeys were available for all drugs we tested except diazepam, phenobarbital, nefopam, and phenytoin. Results in the two species agreed for 24 of the 27 drugs compared. The exceptions were nalorphine and ethylketazocine (inactive in monkeys but active in rats) and ethanol (active in monkeys but inactive in rats). Cyclazocine, naloxone, caffeine, imipramine, chlorpromazine, and haloperidol were inactive in both species, and the remaining 16 drugs were active in both species.

Screening for reinforcing activity of psychoactive drugs in rats gives, with few exceptions, results which parallel those obtained with monkeys. The use of rats makes practical the evaluation of new psychoactive drugs for potential abuse early in their development.


Collins, R. J., Weeks, J. R., Cooper, M. M., Good, P. I., & Russell, R. R. (1984). Prediction of abuse liability of drugs using intravenous self-administration by rats. Psychopharmacology, 82, 6-13.

Griffiths, R. R., & Balster, R. L. (1979). Opioids: Similarity between evaluation of subjective effects and animal self-administration results. Clinical Pharmacology and Therapeutics, 25, 611-617.

Griffiths, R. R., Bigelow, G. E., & Henningfield, 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.

Griffiths, R. R., Brady, J. V., & Bradford, L. D. (1979). Predicting the abuse liability of drugs with animal drug self-administration procedures: Psychomotor stimulants and hallucinogens. In T. Thompson & P. B. Dews (Eds.), Advances in behavioral pharmacology (Vol. 2, pp. 163-208). New York: Academic Press.

Johanson, C. W., & Balster, R. L. (1978). A summary of the results of a drug self-administration study using substitution procedures in rhesus monkeys. Bulletin of Narcotics, 20, 43-54.

Morrison, D. F. (1967). Multivariate statistical methods. San Francisco: McGraw-Hill.

van Ree, J. M. (1979). Reinforcing stimulus properties of drugs. Neuropharmacology, 18, 963-969.

Schuster, C. W., & Thompson, T. (1969). Self-administration of and behavioral dependence on drugs. Annual Review of Pharmacology, 9, 483-502.

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

Weeks, J. R. (1972). Long-term intravenous infusion. In R. D. Myers (Ed.), Methods in psychobiology (Vol. 2, pp. 155-168). London: Academic Press.

Weeks, J. R. (1977). The pneumatic syringe: A simple apparatus for self-administration of drugs by rats. Pharmacology Biochemistry & Behavior, 7, 559-562.

Weeks, J. R. (1979). A method for administration of prolonged intravenous infusion of prostacyclin (PGI ) to unanesthetized rats. Prostaglandins, 17, 495-499.

Weeks, J. R. (1981). An improved pneumatic syringe for self-administration of drugs by rats. Pharmacology Biochemistry & Behavior, 14, 573-574.

Weeks, J. R., & Collins, R. J. (1976). Changes in morphine self-administration induced by prostaglandin E and naloxone. Prostaglandins, 12, 11-19.

Weeks, J. R., & Collins, R. J. (1979). Dose and physical dependence as factors in the self-administration of morphine by rats. Psychopharmacology, 65, 171-177.

Woods, J. H., Smith, C. B., Medzihradsky, F., & Swain, H. H. (1979). Preclinical testing of new analgesic drugs. In R. E. Beers & E. G. Bassett (Eds.), Mechanisms of pain and analgesic compounds, (pp. 429-445). New York: Raven Press.

Yanagita, T. (1977). Brief review on the use of self-administration techniques for predicting drug dependence potential. In T. Thompson & K. R. Unna (Eds.), Predicting dependence liability of stimulant and depressant drugs, (pp. 231-242). Baltimore: University Park Press.


The screening test described above is based upon the activity of the drug, i.e., the dose required to reinforce self-administration. The strength of such a reinforcement is another factor in abuse liability. The latter has been evaluated in monkeys using progressive-ratio programs. Since this chapter was submitted, we have completed preliminary studies evaluating the rat in a progressive-ratio program. Groups of naive rats were offered morphine under a continuous reinforcement schedule for 5 days, then on a fixed ratio schedule increased daily to the break point (less than 4 injections in one day). Results were expressed as the mean of the daily steps to the breaking point. A progressively decreasing geometric progression was used, the initial increment being log 0.20 (1.58 fold), decreasing 4.5 percent each step. This decreasing increment prevented excessively large (arithmetic) increases in the work requirement, which might compromise discrimination between different strength reinforcements at higher ratios. Thus, in this series, steps 13 and 20 are FR-100 and FR-473, respectively, but without the decreasing increment they would have been FR-398 and FR-10,000.

Results for morphine are summarized in the table. The mean breaking points are linearly related to the log dose (b = 4.4, r = 0.69). However, the protocol used may not be the most suitable. It is time-consuming, and at the lower fixed ratios some rats may consume toxic quantities of drug. J. E. Moreton (personal communication) has suggested that instead of a daily increase in fixed ratios, the ratio be increased after a set number of reinforcements. Likewise, the 5-day continuous reinforcement training period could be amended to allow starting the progressive-ratio program when the daily injection rate exceeded a specified number of injections. These preliminary studies suggest that the rat is a suitable species for progressive-ratio studies on reinforcing drugs.

Table 1
The Relative Strength of Reinforcement of Different Doses of
Morphine Sulfate Using the Progressive-Ratio Method
of Rats
Steps to Breaking Point
Mean ± S.E.
Approximate FR
(by interpolation)
0.032 10 8.0 ± 1.32 23
0.10 10 11.1 ± 1.53 60
0.32 10 12.7 ± 1.07 92
1.0 10 15.0 ± 0.39 165
3.2 10 17.1 ± 0.82 265

©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 05 April 2009 11:40 EDT.
Send comments to:
Report technical problems to: