Many of the methods for assessing drug reward involve examining the ability of a drug to serve as a reinforcer. Frequently, some response is performed by the subject followed by presentation of the drug reward. This approach evaluates the ability of a drug to directly reinforce behavior and is a simple extension of operant psychology. Presentation of the reinforcing stimulus (i.e., rewarding drug) is contingent upon the performance of some behavioral response such as lever pressing. Thus, the drug’s capacity to directly control behavior is assessed. An alternative approach involves studying the associations developed between a rewarding drug effect and some arbitrarily selected stimulus. This conditioning approach does not involve the subject "working" for the drug reward, but rather, it examines the behavior of the subject following presentation of stimuli associated with drug reward.

     Perhaps the earliest study demonstrating that stimuli associated with drug presentation can elicit behaviors described as drug seeking is Spragg’s (1940) study of "Morphine Addiction in Chimpanzees." In this study chimpanzees were injected daily with morphine and Spragg reported signs of drug-seeking behavior when the subjects were brought into the room associated with the injections. No attempt was made to quantify these behaviors, however, and the effect was attributed solely to morphine’s ability to relieve withdrawal distress in physically dependent animals. Nonetheless, this study appears to be the first to document drug-seeking behavior in animals and emphasized the importance of stimuli associated with rewarding drug injections in the control of the animals’ behavior.

     Another early study assessing the ability of stimuli associated with drug reward to influence behavior was reported by Beach (1957). Animals received injections of morphine and were "run into" and remained in an environment paired with the drug effect. After this procedure was repeated for several days, subjects were tested for their attraction to the environment associated with the drug effect. During the test trials, the animals showed a marked preference for the compartment associated with the drug effect. This study involved an instrumental response (i.e., running in a Y-maze), but it is properly considered a conditioning study because the subjects were never reinforced for their performance in the maze; the drug effect was simply associated with "running into" and with the stimuli contained in a specific compartment of the apparatus.

     The Beach (1957) study is one of the first demonstrations of drug-reward conditioning, but it involved forcing the animals to "run into" a specific compartment during training. Kumar (1972) reported a study that did not involve an instrumental response associated with the drug effect. In his study, animals were injected with morphine and directly placed in a compartment associated with the drug effect. Thus, the subjects did not run to the drug-associated environment during training. Kumar (1972), however, used very large doses of morphine (i.e., 120 mg/kg) and his data were interpreted as measuring the ability of stimuli associated with the relief of withdrawal distress to serve as a conditioned reinforcer.

     Rossi and Reid (1976) reported that a moderate dose of morphine (i.e., 10 mg/kg) associated with a specific environment produced evidence of conditioning that was related to the drug’s rewarding effects. In their study animals were injected with morphine and later placed directly into a specific compartment. This method was used as an independent assessment of morphine reward and was offered as corroborative evidence of morphine reward at times-post-injection that corresponded to morphine’s facilitation of brain stimulation reward (see Reid, this volume). They further suggested that the conditioned place preference method was a measure of the drug’s "affective consequences" and emphasized the positive hedonic impact of morphine administration. This appears to be the first study to employ place preference conditioning in its contemporary fashion. Two factors distinguish their study from the earlier work: (i) no instrumental response was associated with drug conditioning and (ii) a relatively low dose of morphine was used. Considering the current popularity of the conditioned place preference method, it is interesting to note that Rossi and Reid’s application of place preference conditioning was not primarily designed to develop a new method of assessing drug reward, but rather, it was designed to provide independent validation of another method (i.e., brain stimulation reward) of assessing the rewarding effects of morphine.

     During the past several years, numerous studies have reported evidence of drug reward using place preference conditioning. Because this procedure appears to be relatively simple, it has generated a considerable amount of excitement in the study of drug reward. Surgical preparation of the subjects is not necessary and minimal equipment is required. Data collection can be easily automated using a microcomputer (Bozarth, 1983), and drug testing can be completed in a few days. These factors allow the rapid screening of large numbers of compounds with very little effort. Despite the widespread appeal of this technique, there is controversy regarding the validity of some place preference procedures. Furthermore, the characteristics of this phenomenon have not been systematically examined and are poorly understood.

     The purpose of this series of experiments was to explore some of the variables likely to influence place preference conditioning. The merit of this procedure has relied primarily on its face validity: place preference is an intuitively satisfying statement about the relationship between a rewarding event and stimuli associated with that reward. To assess the validity of this method, it is necessary to characterize the influence of (i) variables known to affect drug reward and (ii) variables known to govern the strength of conditioning. This paper explores the effects of manipulating some variables that have well established effects on classical conditioning. If conditioned place preference studies truly represent a class of behavior governed by the association of drug reward with environmental stimuli, then manipulation of variables known to affect conditioning should influence the development of conditioned place preference. This is one approach to empirically validating conditioned place preference studies. Without such empirical validation, the interpretation of data generated from this technique is questionable.

General Experimental Procedure

     Most of the experiments reported in this chapter use the same general procedure. Subjects were experimentally naive, male, Long-Evans rats usually weighing 300 to 400 grams. They were individually housed in cages with a wire mesh bottom. Food and water were available ad libitum, except during conditioning and behavioral testing. A 12-hour light/dark cycle was used, and testing occurred during the light phase of this cycle. The apparatus consisted of a shuttle box measuring 25 x 35 x 25 cm with a smooth, Plexiglas floor on one end of the chamber and a tubular stainless steel floor on the other end. The amount of time spent on the smooth side of the apparatus was automatically recorded as were the number of crosses between the two sides (Bozarth, 1983).

     The standard procedure usually consisted of three phases. The initial place preferences of the subjects were recorded for five 15-minute trials during the first week. The last trial of this series served as a measure of the animals’ preconditioning place preferences. Next, a barrier was inserted in the shuttle box dividing it into two compartments. The subjects were subcutaneously injected with drug and immediately placed in one compartment of the apparatus (usually their nonpreferred side) for 30 minutes. After three such conditioning trials, place preference was again measured when the subjects were allowed free access to the entire test apparatus. Each subject’s change in place preference was computed by subtracting the place preference recorded during the last preconditioning trial from that of the test trial. These data were then grouped according to treatment conditions and the appropriate statistical analysis performed.

     There are several points that should be noted with this procedure. First, most subjects (c. 80%) showed a strong preference (c. 70% test duration) for the tubular side of the shuttle box. The reason for this is unclear, but it may be related to the fact that they were housed in cages with a wire mesh floor. Second, subjects were usually conditioned on their nonpreferred sides of the test apparatus. This was done to maximize the potential shift in place preference following conditioning with appetitive rewards. If the objective of the study were to assess place aversions, conditioning on the preferred side would be required to avoid limiting the potential magnitude of the shift. Third, the shuttle boxes were thoroughly washed before each phase of the procedure--prior to preconditioning, conditioning, and test trials. Although olfactory cues are extremely important to rats, this procedure is necessary when large numbers of subjects are tested in the same boxes. Otherwise, the conditioning of one subject may affect the measurement of preference in another; a rat may be responding to the olfactory cues left by earlier rats rather than making a place preference discrimination based on its own conditioning experience.

Reliability of Three-Trial Conditioning

     One of the first considerations in the utility of conditioned place preference studies is the reliability of the conditioned effect. Figure 1 illustrates eight independent studies using the same basic conditioning procedure. All subjects, except those in Group A, were subcutaneously injected with 0.3 mg/kg of heroin during three 30-minute conditioning trials. Subjects in Group A were injected with 0.5 mg/kg of heroin during four conditioning trials. Other subjects were injected with physiological saline (1 ml/kg) during their conditioning trials. A single 15-minute test trial measured the place preference following conditioning.
 

Figure 1: Reliability of the conditioned place preference produced by heroin using a standard conditioning procedure. The various experiments were independently conducted over a 4-year period (n = 11 to 15/group). The figure shows the mean ± SEM) change in place preference following conditioning. Striped bars, heroin conditioned; open bars, saline conditioned.

     An analysis of variance (ANOVA) performed on the data from Groups B through H showed that conditioning with heroin reliably produced a conditioned place preference [F (1,161) = 54.940, p < 0.001], while saline conditioning resulted in only sporadic shifts in place preference. There was no significant difference across the various replications [F (6,161) = 1.111, p > 0.1], nor was there any Treatment x Replication interaction [F (6,161) = 0.407, p > 0.1].

Statistical Considerations

     It is interesting to note that of the several ways that these data could be statistically analyzed, Fisher’s (1935) method of specific comparison is perhaps the most appropriate. A series of dependent t-tests can be used to assess shifts in individual groups with the alpha level adjusted to protect against the use of multiple t-tests (i.e., cumulative error rate). By setting a* = a/r where r = the number of tests, a* protects against the influence of repeated tests. The use of this method has the advantage of minimizing the effects of unequal numbers of subjects and differences in the homogeneity of variances among treatment conditions. The practical value of this approach can be seen in Table 1 which shows that all of the heroin-conditioned and none of the saline-conditioned groups demonstrated a reliable change in place preference.

     For the seven t-tests between groups, a* = 0.007. Two of the seven groups failed to exceed this critical value (i.e., Groups C and E). The within group comparisons consist of 15 individual tests; for this series, a* = 0.003. All drug-treated groups except Group G exceeded this criterion; none of the saline groups even approached this critical value of a*. Both the t-test for simple main effects (Winer, 1971) and the Tukey’s (a) test revealed between group differences for all comparisons.

     Although the use of repeated t-tests is very prone to produce Type II errors and some adjustment of the a-level is appropriate, some statisticians (e.g., Lindman, 1974) suggest that the adjusted a-level can be determined on logical grounds and need not rely strictly on statistical approaches. If a is set to 0.005 for each of the 15 specific comparisons across all groups, all of the drug-treated groups show a reliable conditioned place preference while none of the saline-treated groups show significant shifts. This adjustment of retains much of the power of individual t-tests while minimizing the probability of a Type II error. Furthermore, the results of this approach are very similar to those of both the t-test for simple main effects based on the ANOVA and the more conservative Tukey’s (a) test. Thus it would appear that a = 0.005 is a suitable criterion for determining the significance of shifts in place preference, and this approach offers the advantage of minimizing the influence of large, within group variance that sometimes occurs with various treatments. In fact, if an a-level of 0.05 were used to assess changes in place preference, only one saline-conditioned group would have shown a significant shift in place preference (false positive), and even this shift in place preference could have resulted from reward-induced conditioning (see section on Stress-Induced Place Preference).
 

     With the ANOVA approach, an extremely high variance in one treatment cell contributes disproportionately to the overall within-cell error variance. Small differences between other cells may be masked, even though their within-cell variances are very small. With the dependent t-test approach, each group is assessed independently for changes in place preference. Extremely high within-cell variance in one group will only affect the statistical outcome of that group and not the others involved in the comparisons. Thus, if the a-level is adjusted to minimize the probability of a Type II error, the use of individual dependent t-tests can greatly enhance the power of the statistical analysis.

     Another statistic that is useful in describing the relationship between the experimental manipulation and the observed effect is strength of association measures. For t-tests, h² is the appropriate measure (see Linton & Gallo, 1975), and the values of this statistic for each of the comparisons are shown in Table 1. The mean (± SEM) proportion of variance associated with treatment across all groups was 67.6% (± 3.5%) for drug treatment, 9.2% (± 5.4%) for saline treatment, and the mean strength of association between groups was 26.4% (± 3.2%). The mean strength of association measure based on the between groups t-tests is very close to the strength of association measure based on the ANOVA (w² = 24%; see Linton & Gallo, 1975; Winer, 1971). Although the strength of association measure for the between groups comparisons may not appear to be very large, Linton and Gallo (1975) suggest that most studies fail to account for more than 10% of the variance due to treatment effects. Thus, the overall strength of association for the between groups differences is acceptable.

Stability of Preconditioning Preferences

     One factor to consider when assessing the validity of conditioned place preference studies is the stability of the subjects’ preconditioning preferences. If these scores were not stable, then shifts in preference might not accurately reflect appetitive conditioning; but, rather, they might result from spontaneous changes in the preconditioning measures. Although the data presented in Figure 1 would suggest that such spontaneous shifts are not significant, an examination of the stability of these scores would help to define the characteristics of place preference testing.

     To assess the stability of these preconditioning place preferences, a group of 20 animals was tested for 15 minutes per day for four 5-day blocks of testing. These subjects had free access to the entire shuttle box, and testing was on consecutive 5-day intervals with two days of no testing intervening between each block. Figure 2 reveals that the measure of preconditioning place preference displayed little variation across repeated testing. A similar effect was noted for the measure of locomotor activity (i.e., crosses, data not shown). The mean amount of time spent on the tubular side of the test apparatus was much higher than that spent on the smooth side. An examination of the preconditioning preferences across a large number of groups (i.e., the last trial of the 5-day preconditioning sequence) revealed that this preference is consistent across different groups of experimental subjects tested during the past several years (data not shown).
 

Figure 2: Stability of preconditioning place preference across 20 days of testing. Animals were allowed free access to the test apparatus for 15 minutes per day. The graph shows the mean (± SEM) time spent on the smooth side of the test apparatus.

Dose-Response Analysis

     One of the factors that has been shown to influence the strength of a conditioned response is the magnitude of the unconditioned stimulus. In general, increasing the dose of drug tested results in an increase in the behavioral response produced. Following 5 days of preconditioning trials, several groups of rats (n = 19/group) were injected with heroin (0.03 to 1.0 mg/kg) and conditioned for three 30-minute conditioning trials. Various doses of heroin were tested to determine if a graded response might be produced. The data were analyzed by subtracting each subject’s preconditioning score from the time spent on the conditioning side during the test trial.

     Figure 3 illustrates the influence of drug dosage on the magnitude of the conditioned place preference. The two lowest doses failed to produce a significant place preference as shown by individual t-tests with a* = 0.01 [t’s (18) = 1.113 & 2.229, p’s < 0.5 & 0.05, respectively]. The remaining three doses all produced significant shifts in place preference [t’s (18) = 4.153 to 5.356, p’s < 0.001]. The strength of association measures (computed from the individual dependent t-tests) were: 7%, 21.6%, 48.9%, 59.2%, and 61.4% in ascending dose order. As can be seen in the figure, there is a shallow, but statistically significant, dose-response effect [F (4,90) = 4.592, p < 0.005] with little evidence of conditioning produced by 0.03 mg/kg and maximum conditioning produced by 0.3 mg/kg. These data suggest that the conditioning is sensitive to drug dosage, but that there is relatively little difference between the threshold response and maximum response using this conditioning protocol.
 

Figure 3: Dose-response analysis of heroin-induced conditioned place preference. The figure shows the mean (± SEM) shifts in place preference after conditioning with various doses of heroin.

Influence of Number of Conditioning Trials

     Another variable that has been shown to influence the strength of a conditioned response is the number of stimulus-response pairings. To determine if the number of conditioning trials affected the subsequent conditioned place preference, rats (n = 11 to 12/group) were injected with heroin (0.3 mg/kg) and conditioned for 30 minutes as in the previous experiments. Three groups of subjects were conditioned for one, three, or ten such conditioning trials. Other animals were injected with saline (1 ml/kg) and conditioned for the same number of trials to determine if habituation to the nonpreferred side would affect the measure of place preference.

     As with the dose-response analysis, increasing the number of conditioning trials seemed to have little effect on the measure of conditioned place preference [Drug: F (1,66) = 28.221, p < 0.001; Trials, F (2,66) = 1.424, p > 0.1]. Despite the failure to find a significant effect for the factor of Number of Trials or a significant Drug x Trial interaction [F (2,66) = 0.808, p > 0.1], subjects conditioned for three trials appeared to show a stronger conditioned place preference than subjects receiving only one conditioning trial. This impression is supported by a comparison of the strength of association measures (based on the dependent t-tests) for the one- and three-trial conditioning durations (48% and 65%, respectively). Extending the number of conditioning trials to ten, however, clearly failed to produce a stronger conditioned response (strength of association = 65%).
 

Figure 4: Effect of the number of conditioning trials on the strength of the conditioned place preference. The figure shows the mean (± SEM) change in place preference. Striped bars, heroin conditioned; open bars, saline conditioned.

Influence of Conditioning Trial Duration

     The duration of exposure to the unconditioned stimulus is another variable that might influence the magnitude of conditioning seen with this procedure. Longer or shorter conditioning trials might alter the strength of the conditioned place preference. To test this hypothesis, rats (n = 15/group) were injected with heroin (0.3 mg/kg) and conditioned for either 10, 30, or 100 minutes for a total of three conditioning trials. Other groups of rats were conditioned for the same length of time with saline (1 ml/kg) injections. This latter procedure should assess the influence of habituation on any changes in place preference seen with varying the duration of the conditioning trials.

     There was an effect for the factor of Drug Treatment [F (1,88) = 6.847, p < 0.05], but neither the factor associated with Trial Duration [F (2,84) = 0.199, p > 0.1] nor the Drug x Duration interaction [F (2,84) = 1.222, p > 0.1] was significant. Subjects conditioned for 10 and 30 minutes showed comparable magnitudes of change in place preference. The rats conditioned for 30 minutes, however, were more uniform in their place preferences as reflected by the variances associated with these two conditioning durations (see Figure 5). Subjects conditioned for 100 minutes appeared to display less preference than subjects conditioned for the two shorter periods of time. This is reflected in the strength of association measures for these three trial durations: 49%, 82%, and 39%, respectively. It is likely that the rewarding drug effect did not last for the entire 100-minute period and that exposure to the conditioning environment without this drug action attenuated the conditioned place preference. Alternatively, some aversive effect may accompany the termination of the rewarding drug effect, and this might decrease the net place preference seen with 100-minute conditioning periods. Saline-conditioned rats showed no systematic changes in their place preferences as a function of the duration of the conditioning trials. Thus, habituation to the test apparatus is not likely to be an important factor influencing the animal’s place preference.
 

Figure 5: Influence of the duration of individual conditioning trials on the strength of the conditioned response. The figure shows the mean (± SEM) change in place preference. Striped bars, heroin conditioned; open bars, saline conditioned.

Effect of Pre-exposure to the Conditioned Stimulus

     For a stimulus to serve as an effective conditioned stimulus, it is important that this stimulus be salient during the conditioning trials. Pre-exposure to the test apparatus, which occurs during the preconditioning trials, might decrease the effectiveness of the conditioning environment in serving as a conditioned stimulus. As in the previous studies, rats (n = 7 to 8/group) were injected with heroin (0.3 mg/kg) and conditioned using the standard three-trial conditioning procedure. Other rats were conditioned with saline (1 ml/kg). In this experiment, however, no preconditioning trials were conducted so that the subjects’ first exposure to the conditioning environment occurred with the initial presentation of the unconditioned stimulus (i.e., heroin or saline injections).

     Figure 6 illustrates the place preferences for rats conditioned with heroin and with saline. Because the preconditioning place preferences were not determined, the data represent the time spent on the conditioning side rather than shifts in place preference. Furthermore, the data are illustrated separately for subjects conditioned on the smooth side of the apparatus (i.e., normally nonpreferred side) and for subjects conditioned on the tubular side of the apparatus (i.e., normally preferred side). For both the heroin- and saline-conditioned subjects, there was a strong preference for the tubular side of the apparatus following conditioning trials [F (1,26) = 169.253, p < 0.001]. A significant change in place preference developed in rats conditioned on what would normally be their nonpreferred side of the test apparatus, while conditioning on the normally preferred side was not effective in altering place preference. The combined results of this experiment suggest that an overall shift in place preference was produced in rats with no prior exposure to the conditioning environment [F (1,26) = 12.863, p < 0.005]. There was a significant Drug x Side interaction [F (1,26) = 13.641, p < 0.005].
 

Figure 6: Conditioned place preference seen with no pre-exposure to the test apparatus. Subjects were conditioned using the standard procedure, but no preconditioning trials were conducted. The figure shows the mean (± SEM) time spent on the conditioning side. Striped bars, heroin conditioned; open bars, saline conditioned.

     The magnitude of the change in preference seen in this experiment appeared somewhat greater than that produced with the usual testing procedure. Figure 1 shows that the standard conditioning procedure produced changes in place preferences ranging from 100 to 200 seconds greater than saline conditioning. Without pre-exposure, heroin-conditioned subjects spent about 250 seconds more on the conditioning side than did saline-conditioned subjects (see Figure 6, combined data). This suggests that a slight increase in the shift in place preference is seen when subjects are not pre-exposed to the conditioning apparatus.

Type of Conditioning

     The results of the dose-response analysis and of the experiments concerning the number of conditioning trials, the duration of individual conditioning trials, and conditioning without pre-exposure to the test apparatus were rather disappointing. Although there was some indication of an effect for these manipulations (as would be predicted based on data from classical conditioning experiments), these procedures failed to markedly influence the maximum change in place preference demonstrable with this technique. In fact, ANOVAs failed to reveal significant effects for most of these experimental manipulations, and effects were shown only as changes in strength of association measures. Thus, the data suggest that conditioned place preference studies produce a weak, albeit reliable, index of drug reward.

     Animals conditioned with heroin in the previous experiments only had experience with drug in the conditioning side of the apparatus. It might be that contrasting the drug conditioning with experience in the other compartment after saline injections would enhance the preference for the side associated with the drug effect. This discrimination training was investigated in two ways. One group (n = 11) was injected with saline (1 ml/kg), placed on one side of the shuttle box for 30 minutes and then injected with heroin (0.3 mg/kg) and placed on the other side for another 30 minutes. This procedure was repeated once a day for a total of three days. Another group of subjects (n = 12) received heroin (0.3 mg/kg) in the conditioning side of the apparatus for 30 minutes on one day and was injected with saline (1 ml/kg) and placed in the other side for 30 minutes on alternate days. This procedure was repeated three times so that the number of conditioning trials with heroin was equal to that used in the previous experiments. Groups conditioned with saline in both compartments, either within (n = 12) or between (n = 9) sessions, were also tested using this same protocol as were groups under the standard three-trial, drug-only conditioning procedure (n = 14/group). Other groups (n = 15/group) received heroin (0.3 mg/kg) or saline (1 ml/kg) and were allowed free access to both compartments for 30 minutes during each of three conditioning trials.

     Significant changes in place preference were seen in both groups receiving heroin and conditioned with discrimination training [Contrast-1, t (11) = 4.493, p < 0.001; Contrast-2, t (10) = 4.939, p < 0.001], while saline injections did not significantly modify place preferences seen with this conditioning procedure [Contrast-1, t (11) = 1.333, p > 0.1; Contrast-2, t (8) = 1.677, p > 0.05]. Nonassociative conditioning did not produce reliable shifts in either the heroin-treated [t (14) = 1.033, p > 0.1] or the saline-treated [t (14) = 1.181, p > 0.1] groups. The group conditioned with heroin using the standard procedure also showed a significant shift in place preference [t (13) = 4.874, p < 0.001] while the subjects conditioned with saline did not significantly change their place preference [t (13) = 0.973, p > 0.1].

     Contrasting the effects of saline with those of heroin did not result in a greater change in preference for either the group receiving saline and heroin conditioning on alternate days or for the group receiving saline and heroin conditioning on the same days. This was surprising because experience with saline in one compartment and heroin in the other would be expected to enhance the contrast, and subsequent preference, for the two compartments. Experiments in classical and operant conditioning have shown that stimulus discrimination can be enhanced by training with similar, but perceptually distinct, stimuli.
 

Figure 7: Effect of type of conditioning on the magnitude of the conditioned place preference. Nonassociative conditioning represents data from subjects that were not restricted to one compartment in the conditioning apparatus (i.e., were allowed to move freely in both the smooth and tubular sides). Contrast-1 represents subjects that received discrimination training with heroin- and saline-conditioning occurring on the same day. Contrast-2 represents animals that received discrimination training with heroin- and saline-conditioning occurring on alternate days. Drug-only conditioning represents the standard conditioning procedure where subjects did not receive discrimination training. Striped bars, heroin conditioned; open bars, saline conditioned.

Use of an Olfactory Cue

Increasing the salient cues on the conditioning side might be expected to enhance the conditioning seen with this procedure. Because rats are particularly responsive to olfactory cues, a distinctive odor was paired with one side of the apparatus (i.e., the smooth side). Preconditioning place preferences were measured, and rats subsequently conditioned with heroin (0.3 mg/kg, n = 15) or saline (1 ml/kg, n = 14) using the standard conditioning procedure. In addition to the textural cues normally present, three drops of 10% acetic acid were placed on the extreme end of the conditioning side of the apparatus. The odor cue was present throughout all phases of the experiment--preconditioning, conditioning, and testing.

     Reliable shifts in place preference were seen following conditioning with heroin [t (14) = 4.689, p < 0.001] but not following conditioning with saline [t (13) = 0.976, p > 0.1]. The magnitude of the preference change, however, did not appear to be affected by the addition of the olfactory cue. Also, it is interesting to note that the acetic acid cue had little influence on the preconditioning place preferences in these two groups of rats; the subjects spent about 27% of the time on the side associated with the olfactory cue compared with a mean of around 30% of the time for most experiments that did not have an olfactory cue.
 

Figure 8: Changes in place preference associated with both textural and olfactory cues. Animals were conditioned using the standard procedure except for the addition of an acetic acid odor on the conditioning side. Striped bars, heroin conditioned; open bars, saline conditioned.

Temporal Analysis of Place Preference

     The failure to increase the conditioned change in place preference beyond that produced by the initial conditioning procedure was very surprising. One possible explanation is that the subjects display their maximum change in place preference during the initial minutes of testing and that the remaining test period serves to extinguish the conditioned change in place preferences. That is, the subjects may show a strong conditioned place preference during the first few minutes of testing followed by a return to the preconditioning preferences for the test apparatus. The net place preference would thus include both an initially strong manifestation of conditioning and a later return to the subjects’ preconditioning preferences; the average would yield only a modest overall shift for the total test period. To test this possibility, subjects (n = 20/group) were conditioned with heroin (0.3 mg/kg) or saline (1 ml/kg) for 30 minutes during three conditioning trials. During the test trial, preferences were measured every minute for the duration of the usual 15-minute test period. This temporal analysis of place preference should reveal any decline in preference for the conditioning side as a function of response extinction.

     There was a significant effect for Drug Treatment [F (1,570) = 38.221, p < 0.001] with no significant effect associated with Test Period [F (14,570) = 0.72, p > 0.1] or any Treatment x Period interaction [F (14,570) = 0.829, p > 0.1]. The change in place preference produced by heroin-conditioning showed little change over the first ten minutes of testing (see Figure 9). There was no indication that the subjects spent the majority of time on the conditioning side during the first few minutes of testing. The slight decrease in conditioned place preference seen for the last 5 minutes of testing suggests that tests limited to 10 rather than the usual 15-minute period might reveal a somewhat stronger conditioning effect. However, this is unlikely to have a major influence on the strength of the measured shift in place preference.
 

Figure 9: Changes in place preference within a single 15-minute test period. Each data point depicts the mean (± SEM) time spent on the conditioning side for consecutive 60-second intervals. Circles represent heroin-conditioned subjects, and squares represent saline-conditioned subjects.

Alternative Explanations for Shifts in "Place Preference"

     The fact that manipulations that influence classical conditioning have little or no effect on the conditioned place preference questions the basis of this phenomenon. Although the dose of drug and number of conditioning trials did affect place preference, these effects were minimal and other variables might account for the apparent shift in place preference following repeated injections of heroin.

Regression Toward the Mean

     One factor that might be involved in the increased time spent on the side of putative conditioning is regression toward the mean. Because subjects are usually conditioned on their nonpreferred sides of the test apparatus, spontaneous shifts in the measure of place preference might occur during subsequent testing. An examination of the stability of the preconditioning preferences, however, makes this an unlikely explanation of the shifts in preference. Subjects demonstrated stable preconditioning preferences for 20 days of repeated testing (see Figure 2). Also, animals receiving saline during the conditioning trials failed to show a reliable shift in place preference (see Figure 1) further indicating that spontaneous shifts in preference are not an adequate explanation of this effect.

Habituation or Anxiolytic Drug Action

     The strong preference displayed for one side of the test apparatus during the preconditioning trials might also reflect an aversion to the other side. It could be the case that novelty or some other factor inhibits the exploration of the smooth side of the chamber and that forced habituation or the experience of anxiolytic drug effects might decrease the initial aversion to that side. This is consistent with the observation that animals rarely spend over 50% of the time on the smooth side even after heroin conditioning.

     The results of the experiment that assessed the influence of the number of conditioning trials is relevant to this alternative explanation of conditioned place preference. Note that in Figure 4, increasing the number of conditioning trials from one to three and to ten did not increase the amount of time saline-conditioned subjects spent on the nonpreferred side. This suggests that forced habituation cannot account for the shift in place preference seen following heroin conditioning. Anxiolytic drug effects would also seem unlikely to account for the observed shifts in place preference because at least a modest decrease in anxiety would be expected following the forced habituation trials. Also, a recent study has shown that diazepam (which has only equivocal rewarding properties, see de Wit & Johanson, this volume; Weeks & Collins, this volume) does not produce a shift in place preference (G. Carr, personal communication, but see Spyraki, Kazandjian, & Varonos, 1985). Because diazepam is a potent anxiolytic agent, this observation appears to directly eliminate anxiolytic effects as a possible explanation of place preference conditioning.

Locomotor Activity

     Another potential explanation for the shift in place preference is also related to the strong preconditioning preferences shown by subjects with this procedure. Because the animals spend very little time on the side which will be subsequently used for conditioning and because the change in preference following conditioning seldom results in subjects spending over half of the test trial on the side of putative conditioning, any increase in general activity might be expected to increase the amount of time spent on the nonpreferred side of the apparatus. That is, if the animals increase their locomotor activity, this might result in their spending more time on the nonpreferred side of the shuttle box due to simple exploration.

     One way to test this possibility is to increase the subjects’ locomotor activity and to measure the amount of time spent on the usual side of conditioning (i.e., nonpreferred side). Subjects (n = 18/group) were given five preconditioning trials, and the last trial served as a measure of the subjects’ place preferences. Next, they were injected with either saline (1 ml/kg) or amphetamine (1.0 or 3.0 mg/kg, i.p., 20 minutes prior to testing), and shuttle activity was measured along with the amount of time spent on the smooth side of the test apparatus. The scores derived from this procedure were treated as place preference scores even though the animals never received any conditioning trials and they were tested under the motor-stimulating effect of amphetamine.

     The changes in pseudo-place preference are shown in Figure 10. Amphetamine produced a dose-dependent increase in the number of crosses during the 15-minute test trial [F (2,51) = 17.639, p < 0.001]; slight, but statistically significant, increases in the amount of time spent on the nonpreferred side of the apparatus were also noted [t’s (17) = 2.447 & 2.465, p’s < 0.025], although this effect was not dose-dependent [F (2,51) = 1.889, p > 0.1]. This suggests that increased activity can influence measures of place preference, but this effect was weak in the present experiment (strength of association = 26%). Nonetheless, the potentially confounding influence of locomotor activity on the measurement of place preference illustrates the importance of assessing changes in locomotor activity during testing.
 

Figure 10: Changes in "pseudo-preference" associated with increases in locomotor activity. The panel on the left shows the changes in locomotor activity produced by amphetamine (dA) injections, while the panel on the right depicts changes in the amount of time spent on the nonpreferred side of the test apparatus during the period of increased locomotor activity. Large increases in locomotor activity produce marginal shifts in place preference measurement, but the magnitude of this effect is too weak to account for the conditioned place preference seen in most studies. Also, an analysis of locomotor scores obtained from subjects tested in the previous experiments shows only occasional increases in locomotor activity, with the majority of place preferences produced without any concomitant change in motor activity. One group that did show a significant increase in locomotor activity was the group receiving 0.3 mg/kg of heroin in the dose-response analysis (see Figure 3). This group had a mean (± SEM) increase of 7.1 (± 1.32) crosses during the 15-minute test period. Although this effect was significant [t (18) = 5.379, p < 0.001], an analysis of the relationship between the increased locomotor activity and the increased time spent on the side of conditioning revealed no significant correlation [r (18) = .105, p > 0.05]. Thus, even in cases where significant changes in locomotor activity do occur, increases in the number of crosses are an unlikely explanation of the increase in time spent on the nonpreferred side of the apparatus.

Cue Reinstatement

     The fact that several alternative explanations of place preference described in the preceding section cannot account for this effect supports the notion that this technique can assess the rewarding action of a drug. The failure to find robust effects for factors known to influence classical conditioning, however, questions the conditioning basis of this phenomenon and emphasizes the importance of identifying the factors responsible for limiting the magnitude of conditioning demonstrable with this technique.

     Some of the procedures described in the preceding sections were designed to make the conditioned response stronger by increasing the saliency of the environmental stimuli associated with the drug reward. Perhaps the most salient stimulus associated with the rewarding drug-effect is the drug cue itself. That is, the internal stimuli associated with the presence of the drug during conditioning may be the most prominent. These stimuli may be directly related to the rewarding properties of the drug or they may be secondary effects that are simply associated with its presence (e.g., autonomic side-effects, changes in thermoregulation). The purpose of the present study was to determine if testing the subjects under the drug condition would increase the magnitude of the conditioned shift in place preference.

     The rats were habituated to the test apparatus for 15 minutes a day for five days. The last day of this series served as a measure of each subject’s initial place preference as in the earlier experiments. Next, they received daily injections of heroin (0.3 mg/kg, n = 12) or saline (1 ml/kg, n = 11) and were forced to remain on their nonpreferred side for 30 minutes during three conditioning trials. On the following day, all subjects were injected with saline and tested for changes in place preference during a single 15-minute trial. One day later, each group was injected with the agent used during conditioning (i.e., either heroin or saline) and place preference was measured again.

     As shown in the previous studies, heroin injections produced a shift in place preference when the subjects were tested following saline injections [F (1,21) = 19.901, p < 0.001]. When tested following heroin injections, the magnitude of place preference was appreciably greater [F (1,21) = 7.981, p < 0.012]. Figure 11 illustrates the place preference seen across the two measurements. Heroin-conditioned animals spent significantly more time on the drug-associated side than the saline-conditioned subjects [F (1,21) = 12.459, p < 0.001]; the factor of Trials was also significant [F (2,42) = 12.960, p < 0.001] as was the Drug x Trial interaction [F (2,42) = 9.958, p < 0.001]. The changes in place preference were not associated with a reliable change in locomotor activity [F’s (1,21) = 0.843 and 1.573, p’s > 0.05]. The subjects spent over half of the test time on the side of conditioning when tested under the drug state but not when tested following saline injections (see Figure 12).

     When the data were analyzed in terms of the number of subjects showing an absolute preference for the side of conditioning (i.e., spending 450 seconds on that side), none of the heroin-conditioned subjects during the preconditioning trial and none of the saline-conditioned subjects on any of their three trials spent over half of the total test time on the side of putative conditioning. In the heroin-conditioned group, an absolute preference was demonstrated in 27% of the subjects when tested in the saline state and in 82% of the subjects when the heroin cue was reinstated [c²(2) = 19.56, p < 0.005].

     This study shows that a true (i.e., absolute) place preference can be produced in subjects tested under the drug cue. This effect parallels human studies showing that craving is most pronounced when drug is believed to be available (Meyer & Mirin, 1979) and animal studies demonstrating a reinstatement of lever-pressing following priming injections of drug (de Wit & Stewart, 1983; Stewart & de Wit, this volume). Although the exact nature of this effect is not clear, it is possible that overshadowing limits the place preference demonstrable in subjects tested in the drug-free state. If a strong association were developed between the rewarding drug effect and some interoceptive drug cue, this might limit the strength of the response elicited by environmental cues alone. Hence, the absence of the most salient cue (i.e., the interoceptive drug cue) might explain some of the apparent anomalies observed with most procedures used for assessing conditioned place preference.
 

Figure 11: Changes in place preference without (TT-1) and with (TT-2) reinstatement of the drug cue during testing. The data depict mean (± SEM) changes from preconditioning place preferences; the animals were conditioned with the usual conditioning procedure. Striped bars, heroin conditioned; open bars, saline conditioned.

Figure 12: Changes in place preference without (TT-1) and with (TT-2) reinstatement of the drug cue during testing. The data represent the mean (± SEM) time spent on the side of conditioning, with a total test duration of 900 seconds. Solid line, heroin conditioned; dashed line, saline conditioned.

     There are a number of other factors that are interesting to examine regarding place preference studies. This section explores some of these factors in an attempt to better understand the basic phenomenon and to help define limitations inherent with this approach to the study of drug reward.

Effect of Home Cage Injections

     It has been suggested that the changes in place preference seen with place preference conditioning studies may be unrelated to associative conditioning. That is, shifts in place preference may not depend on a learned association between specific environmental cues and the rewarding action of a drug. One study (Blander, Hunt, Blair, & Amit, 1984) reported a significant shift in place preference following drug injections that were not paired with the environment used to assess place preference. The possibility that place preferences may significantly shift without specific conditioning is potentially a serious problem with place conditioning studies and merits further consideration.

     The stability of the preconditioning place preferences illustrated in Figure 2 indicates that large spontaneous shifts in place preference are unlikely to occur with repeated testing. The initial place preference of these rats was stable across 20 days of repeated testing. The experiment summarized in Figure 7 shows that drug injections followed by access to the entire test chamber do not significantly affect place preference. If drug injections caused a shift in place preference independent of explicitly pairing the drug effect with a specific environment, then changes in place preference should have occurred in this study.

     Although no attempt has been made to replicate the report of shifts in place preference following rewarding drug injections in an environment other than the one used to measure place preference, a similar condition was tested as part of a control procedure for another study. In this experiment initial place preferences were measured using the standard procedure. Three groups of rats (n = 20/group) then received home cage injections of saline (1 ml/kg), pimozide (0.5 mg/kg), or pimozide (0.5 mg/kg) plus heroin (0.5 mg/kg). After three days of injections, place preferences were remeasured and the data scored in the usual manner even though the drug injections were never paired with the test apparatus. Figure 13 shows the results of this procedure.
 

Figure 13: Effects of home cage injections on place preference. The standard experimental procedure was used, except three daily drug injections were given in the home cage, and the drug effects were never paired with the test apparatus. Abbreviations: Sal, saline; PMZ, pimozide; PMZ & HEROIN, pimozide plus heroin.

     There were no significant shifts in place preference for any of the three groups [t’s (19) = 0.635 to 1.074, p’s > 0.1]. Although this test did not use rewarding drug injections (presuming, of course, that the pimozide injections blocked the heroin reward; see Bozarth & Wise, 1981), drug injections in the home cage failed to modify place preferences. This suggests that neither the handling associated with the injection routine nor a general psychotropic drug action influences the animals’ place preferences, and this further corroborates the conclusion from the earlier experiment showing that nonassociative conditioning is insufficient to explain the shifts in place preference.

Stress-Induced Place Preference

     Across a large number of studies using essentially the same conditioning procedure, saline-treated animals occasionally showed surprisingly large changes in place preference (see Figures 1, 5, & 7). Although these shifts in place preference can occur without any conditioning trials (as shown in Figures 7 & 13), some groups of saline-treated animals that showed particularly notable shifts appeared to be more difficult to handle than animals not showing such shifts. The possibility exists that handling procedures may vary across different groups of rats (or even within a group of rats) and that differences in handling can influence the place preferences seen after conditioning. Because endogenous opioid peptides can be released by stress and these opioids have also been implicated in motivated behavior (e.g., see Reid, this volume), it is possible that saline-treated animals sometimes experience a rewarding effect from the stress-induced release of these peptides. In this experiment, the ability of a mild stressor to produce a change in place preference was assessed. Pretreatment with a narcotic antagonist (i.e., naloxone) during the conditioning trials was used to determine if any stress-induced effects were dependent on an endogenous opioid peptide.

     In this laboratory, animals that are difficult to handle are frequently rocked in a swinging motion just prior to drug injections. This procedure makes the animal docile and facilitates drug injections in active animals. In this experiment preconditioning place preferences were assessed in the usual way for 5 days. Next, three groups of rats (n = 19/group) were rocked in a swinging motion (i.e., gently swinging the arm back and forth) for five repetitions and then injected with saline (1 ml/kg), saline plus naloxone (3 mg/kg), or saline plus naloxone (3 mg/kg) dissolved in sodium metabisulfite (0.1%) which made the solution acidic. A 30-minute conditioning trial immediately followed these injections. Another group of rats (n = 19) were placed in the conditioning compartment without any treatment. After three conditioning trials, place preference was remeasured using a 15-minute trial.

     Figure 14 shows the changes in place preference following conditioning. A slight shift in preference was seen after saline injections [t (18) = 2.701, p < 0.02], and this shift was not present in animals injected with naloxone [t (18) = 0.133, p > 0.8]. Injections of naloxone dissolved in an acidic solution produced a strong place aversion [t (18) = 3.101, p < 0.01]. Animals that were simply placed in the conditioning compartment without "rocking" or injections showed no change in place preference [t (18) = 0.145, p > 0.8]. Although the stress-induced change in place preference is very small, it could explain some of the shifts seen in saline-treated animals during the other experiments and illustrates the potential importance of handling procedures in place preference studies. Furthermore, the fact that this shift in preference is blocked by naloxone suggests that endogenous opioid peptides may be important in this effect. The place aversion produced by injections of an acidic naloxone solution shows that drug vehicle and pH can seriously affect place preference and that the drug-vehicle control condition should always be tested concurrently with the experimental groups.
 

Figure 14: Place preference following a mild stress. Animals were treated with saline (SAL + STRESS), naloxone (NX + STRESS), or naloxone dissolved in an acidic solution (NX*SMBS + STRESS). Other animals were gently placed in the conditioning compartment without any stress or injections (NT).

Place Preference with Visual Cues Only

     The place preference technique used throughout this chapter employs what has been termed an "unbalanced" procedure (see van der Kooy, this volume). The animals show strong preconditioning place preferences and conditioning is usually done on the nonpreferred side of the test apparatus. The potential side bias inherent with this procedure could produce changes in place preference that are dependent on this initial side preference and not related to the rewarding drug effect. Although the alternative explanations that appear most viable (i.e., habituation, anxiolytic drug action, increased locomotor activity) are insufficient to explain the changes in preference that are seen with this technique (see earlier section), the facts that a "true" place preference is usually not seen following drug conditioning and the possibility that an unbalanced procedure could be measuring an effect other than a rewarding drug action make the test of place preference in a balanced test apparatus important.

     The test apparatus was modified so that both sides contained the same type of tubular stainless steel floor. Horizontal black stripes (1 inch wide) were painted on one side of the shuttle box and the other walls remained unpainted. Four groups of rats (n = 10/group) received five 15-minute preconditioning trials and the last trial served as a measure of their preconditioning place preferences. Next, one group received heroin injections (0.3 mg/kg) and were forced to remain on one side of the apparatus for 30 minutes; half of the animals were conditioned on the side with horizontal stripes and half were conditioned on the side without stripes. A second group of rats were treated identically, except saline (1 ml/kg) was injected during the conditioning trials. A third group received discrimination training where saline (1 ml/kg) injections were paired with one side for 30 minutes and followed by heroin (0.3 mg/kg) injections paired with the other side for 30 minutes; half of the rats received saline injections on the side with horizontal stripes and heroin injections on the other side while half of the rats were conditioned with saline injections on the unstriped side and heroin injections on the striped side. A fourth group received conditioning trials with saline on both sides of the apparatus. Two additional groups (n = 10/group) did not receive preconditioning trials and were simply injected with either heroin (0.3 mg/kg) or saline (1 ml/kg) and forced to remain in one compartment for 30 minutes; half of each group was conditioned on the striped side and half on the plain side of the shuttle box.

     The use of identical floor textures on both sides of the apparatus produced a procedure with no strong preconditioning preferences (i.e., initial side preferences were balanced). For subjects that received preconditioning trials, the mean (± SEM) time spent on the striped side increased slightly from 411.2 ± 32.8 seconds out of a total test duration of 900 seconds on the first trial to 531.3 ± 38.2 on the last preconditioning trial. Furthermore, because half of the subjects were conditioned on the striped side and half conditioned on the plain side, the side used for drug conditioning was also balanced. Thus this procedure balanced both the initial side preferences and the side used for drug conditioning.
 

Figure 15: Place preference associated with only visual cues using a "balanced" procedure. Striped bars, heroin conditioned; open bars, saline conditioned.

     Subjects conditioned with heroin using the contrast method [i.e., discrimination training; t (18) = 2.148, p < 0.025] or conditioned without preconditioning trials [t (18) = 2.180, p < 0.025] showed reliable shifts in place preferences. Animals conditioned using the standard conditioning procedure, however, failed to show a significant change in place preference [t (18) = 1.119, p > 0.1; see Figure 15]. The only group to show an absolute place preference (i.e., spending over 450 seconds on the drug-associated side) following conditioning was the group conditioned using discrimination training.

     When the changes in place preferences seen with the standard conditioning procedure were assessed in reference to which side of the test apparatus the subjects received their conditioning trials, a slight shift in place preference was seen when the subjects were conditioned on the striped side but no change in preference was seen when the subjects were conditioned on the plain side (see Figure 16). It appears that this conditioning procedure was not effective when subjects were conditioned in the absence of the strong visual cue associated with the striped end of the shuttle box. When the data from the contrast conditioning procedure were examined as a function of the conditioning side, a conditioned place preference was seen following conditioning on either the striped or plain side of the apparatus (see Figure 17). In this case the subjects had experience with the striped end whether drug conditioning occurred there or in the other compartment. This probably produces a strong discrimination that is reflected by the place preference either toward or away from the striped cue. A similar effect was seen in animals that did not receive preconditioning trials (data not shown).
 

Figure 16: Conditioned place preference associated with only visual cues using the standard (drug only) conditioning procedure. Animals conditioned on the striped side and the plain side are shown separately. Striped bars, heroin conditioned; open bars, saline conditioned.

Figure 17: Changes in place preference associated with visual cues only in animals that received discrimination training. Animals conditioned on the striped side or plain side are shown separately. Striped bars, heroin conditioned; open bars, saline conditioned.

     The results of most experiments reported in this chapter were surprising in several ways. First, they failed to reveal strong effects for factors that have been shown to influence classical conditioning. Although there is some indication of a dose-response function, some influence of the number and duration of conditioning trials, and a small effect of pre-exposure to the conditioned stimulus, these manipulations were ineffective in causing a major change in the magnitude of the conditioned place preference. Second, the initial work in this laboratory reported a conditioned place preference from heroin (Bozarth & Wise, 1981) that was based on somewhat arbitrarily selected parameters. The magnitude of this preference remains the same as that produced by more careful selection of testing parameters. Although the strength of association measures reveal that a reasonable proportion of variance is explained by the experimental manipulation, it is disappointing that few of the procedures examined in this chapter produced a significant increase in this measure. Third, similar strengths of conditioning are produced by various laboratories using markedly different procedures as well as by the different conditioning procedures examined in this chapter. Even with different compounds and different routes of administration (e.g., peripheral vs. central injections), most studies report a shift in place preference around 100 to 200 seconds. This is surprising given the variety of methods and drugs that are used in conditioned place preference studies.

Central Topics

     There has been considerable discussion regarding "the right way and the wrong way" to conduct conditioned place preference studies. Some investigators suggest that they employ the only method of place preference conditioning that works. Although their logic appears sound and the conclusions they reach sensible, little empirical data are offered to support the strong assertions made by these investigators. The series of experiments reported in this chapter systematically examined variables that seemed likely to influence place preference conditioning. None of these manipulations produced dramatic changes in the observed behavior, except the experiment involving reinstatement of the drug cue. Thus, the present data argue that there is no one method of place preference conditioning that is obviously superior to the other methods used by the various laboratories. Nonetheless, a procedure where the initial side preferences and the side of drug-conditioning are fully balanced may be advantageous (e.g., Mucha, van der Kooy, O’Shaughnessy, & Bucenieks, 1982; van der Kooy, this volume), although the lack of proper "balancing" does not invalidate the conclusions based on other (i.e., unbalanced) conditioned place preference procedures.

     Most place preference experiments involve procedures where animals have strong initial preferences for one side of the apparatus. The subjects are usually conditioned on their nonpreferred sides, and a conditioned place preference is reported when a significant increase in the amount of time spent on the drug conditioning side occurs. Even after conditioning, however, the subjects seldom spend over half of the total test time on the conditioning side. Because of this, two controversial issues have emerged. First, it has been argued that the term place preference is a misnomer. Although this observation is correct, the widespread use of the term conditioned place preference to describe increases in the amount of time spent in the compartment associated with drug reward justifies its use de facto. The consistent use of a term to describe an experimental procedure facilitates literature indexing and scientific communication; also, because most investigators are familiar with the lack of an absolute place preference, this term is not generally misleading. Second, some investigators argue that procedures involving strong preconditioning preferences produce a "biased method" of assessing conditioned place preference. Unfortunately, the most common approach to equalizing the amount of time spent by the animals on each side of the apparatus does not really involve removing the "bias" but rather introduces additional biases that result in a more even distribution of time spent in each compartment. For example, other cues (e.g., olfactory, visual, thermal) can be provided that produce a more equal distribution of time spent in each compartment of the apparatus. A potential problem arises if a drug treatment produces a change in reactivity to these cues. This problem could also occur in procedures using a single type of cue, but it is probably more serious in procedures using multiple sensory cues. Another approach to removing strong initial side preferences is to eliminate the stimuli that produce them. The experiment reporting a shift in place preference using only visual cues is one such example (see Figure 15). Perhaps the best solution is to condition some subjects on each side of the apparatus so that half of the subjects are conditioned toward the side of their initial preference while half are conditioned away from the side of their initial preference. This approach may not be successful, however, in procedures with strong initial side preferences (Schenk, Ellison, Hunt, & Amit, 1985) or in procedures with drug-only conditioning (see Figure 16).

Recommendations

     Although conditioned place preference appears to be a reliable measure of the rewarding properties of drugs, caution should be exercised in the use of this technique. Some factors deserve particular attention when conducting place preference experiments: (a) locomotor activity associated with place preference testing should be measured, (b) the effectiveness of a drug treatment known to produce a shift in place preference and the effect of saline should be tested concurrently with the experimental treatments, (c) handling stress should be minimized and equated across all treatment conditions, (d) a priori hypothesis testing and replication of important effects is warranted, and (e) comparisons with other treatment conditions measured at different times should be conservative. As with other measures of drug reward, place preference data are best considered in addition to other lines of evidence supporting a particular hypothesis. Unlike some of the other methods, place preference data are particularly suspect when they are the only supporting evidence for a hypothesis that conflicts with other lines of scientific data. When these and other precautions are observed, conditioned place preference offers an important adjunct to the other methods for assessing drug reward.

     Further work is obviously needed to fully understand this method of assessing drug reward. The conditioned place preference technique, however, can offer an additional method of assessing drug reward, and it may provide the opportunity to ask questions that other methods are not well suited to answer. First, place preference studies provide independent corroboration of reward assessed by other measures. This can strengthen the data base that is used to derive important scientific conclusions. Second, large numbers of compounds can be quickly tested using this technique making it an excellent method for the initial screening of drug reward. Drug dosage and other parameters can be estimated from place preference data prior to testing with other more direct methods such as intravenous self-administration. Third, there are applications where a conditioning measure of drug reward is required. For example, the effect of neuroleptic challenge on intravenous opiate self-administration reveals that animals decrease their drug intake, but it has been suggested that this is caused by a sedative drug action and does not reflect an attenuation of opiate reward (Ettenberg, Pettit, Bloom, & Koob, 1982). Place preference conditioning studies permit testing the effect of neuroleptic challenge of drug reward without the potentially confounding influence of sedative effects. These studies have shown that opiate reward is indeed attenuated by neuroleptic treatment (Bozarth & Wise, 1981; Phillips et al., 1982; Schwartz & Marchock, 1974; cf. van der Kooy, this volume) and they help clarify the interpretation of data from intravenous self-administration studies.

Despite the failure to fully understand the basis of place preference conditioning, this technique clearly discriminates drug reward from nonreward. The ability of place preference conditioning to reliably distinguish the effect of a rewarding drug (i.e., heroin) from saline treatment was illustrated in Figure 1. Furthermore, Table 2 summarizes some of the studies reporting a conditioned place preference from various compounds, and most of these compounds have rewarding properties that have been confirmed with other procedures. Table 3 shows that not all psychoactive compounds produce a conditioned place preference; most of these compounds have not been reported to be rewarding. Although the shift in place preference is usually not large, the procedure appears to be reliable and yields results consistent with other measures of drug reward.
 

     When used with appropriate precautions, conditioned place preference studies can provide an important addition to the other measures used to assess the rewarding properties of drugs. Because the technique is new and not well understood, most conclusions drawn from this paradigm should be conservative and viewed with regard to the results obtained using other measures of drug reward. Nonetheless, the conditioned place preference method offers an important new tool for studying the rewarding effects of abused drugs.
 

     Lydia Alessi is thanked for meticulously conducting the experiments described in this chapter. The research was supported by grants from the Natural Sciences and Engineering Research Council of Canada (NSERC) and by the National Institute on Drug Abuse (U.S.A.). The author is a University Research Fellow sponsored by NSERC.

Advokat, C. (1985). Evidence of place conditioning after chronic intrathecal morphine in rats. Pharmacology Biochemistry & Behavior, 22, 271-277.

Asin, K. E., & Wirtshafter, D. (1985). Clonidine produces a conditioned place preference in rats. Psychopharmacology, 85, 383-385.

Asin, K. E., Wirtshafter, D., & Tabakoff, B. (1985). Failure to establish a conditioned place preference with ethanol in rats. Pharmacology Biochemistry & Behavior, 22, 169-173.

Bardo, M. T., Miller, J. S., & Neisewander, J. L. (1984). Conditioned place preference with morphine: The effect of extinction training on the reinforcing CR. Pharmacology Biochemistry & Behavior, 21, 545-549.

Beach, H. D. (1957). Morphine addiction in rats. Canadian Journal of Psychology, 11, 104-112.

Blander, A., Hunt, T., Blair, R., & Amit, Z. (1984). Conditioned place preference: An evaluation of morphine’s positive reinforcing properties. Psychopharmacology, 84, 124-127.

Bozarth, M. A. (1983). A computer approach to measuring shuttle box activity and conditioned place preference. Brain Research Bulletin, 11, 751-753.

Bozarth, M. A., & Wise, R. A. (1981). Heroin reward is dependent on a dopaminergic substrate. Life Sciences, 29, 1881-1886.

Bozarth, M. A., & Wise, R. A. (1982). Localization of the reward-relevant opiate receptors. In L. S. Harris (ed.), Problems of drug dependence, 1981 (National Institute on Drug Abuse Research Monograph 41, pp. 158-164). Washington, DC: U.S. Government Printing Office.

Bozarth, M. A., & Wise, R. A. (1983). Dissociation of the rewarding and physical dependence-producing properties of morphine. In L. S. Harris (Ed.), Problems of drug dependence, 1982 (National Institute on Drug Abuse Research Monograph 43, pp. 171-177). Washington, DC: U.S. Government Printing Office.

Carr, G. D., & White, N. M. (1983). Conditioned place preference from intra-accumbens but not intra-caudate amphetamine injections. Life Sciences, 33, 2551-2557.

Cunningham, C. L. (1981). Spatial aversion conditioning with ethanol. Pharmacology Biochemistry & Behavior, 14, 263-264.

de Wit, H., & Stewart, J. (1983). Drug reinstatement of heroin-reinforced responding in the rat. Psychopharmacology, 79, 29-31.

Ettenberg, A., Pettit, H. O., Bloom, F. E., & Koob, G. F. (1982). Heroin and cocaine intravenous self-administration in rats: Mediation by separate neural systems. Psychopharmacology, 78, 204-209.

Fisher, R. A. (1935). Cited in H. R. Lindman (1974), Analysis of variance in complex experimental designs. San Francisco: W. H. Freeman.

Fudala, P. J., Teoh, K. W., & Iwamoto, E. T. (1985). Pharmacologic characterization of nicotine-induced conditioned place preference. Pharmacology Biochemistry & Behavior, 22, 237-241.

Gilbert, D., & Cooper, S. J. (1983). Beta-phenylethylamine-, d-amphetamine- and l-amphetamine-induced place preference conditioning in rats. European Journal of Pharmacology, 95, 311-314.

Giovino, A. A., Glimcher, P. W., Mattel, C. A., & Hoebel, B. G. (1983). Phencyclidine (PCP) generates conditioned reinforcement in the nucleus accumbens (ACC) but not in the ventral tegmental area (VTA). Society for Neuroscience Abstracts, 9, 120.

Glimcher, P. W., Giovino, A. A., Margolin, D. H., & Hoebel, B. G. (1984a). Endogenous opiate reward induced by an enkephalinase inhibitor, thiorphan, injected into the ventral midbrain. Behavioral Neuroscience, 98, 262-268.

Glimcher, P. W., Margolin, D. H., Giovino, A. A., & Hoebel, B. G. (1984b). Neurotensin: A new ‘reward peptide.’ Brain Research, 291, 119-124.

Katz, R. J., & Gormezano, G. (1979). A rapid and inexpensive technique for assessing the reinforcing effects of opiate drugs. Pharmacology Biochemistry & Behavior, 11, 231-233.

Kumar, R. (1972). Morphine dependence in rats: Secondary reinforcement from environmental stimuli. Psychopharmacology, 25, 332-338.

Kurz, E. M., & Levitsky, D. A. (1983). Lithium chloride and avoidance of novel places. Behavioral Neuroscience, 97, 445-451.

Lindman, H. R. (1974). Analysis of variance in complex experimental designs. San Francisco: W. H. Freeman.

Linton, M., & Gallo, P. S. (1975). The practical statistician: Simplified handbook of statistics. Monterey, CA: Brooks/Cole Publishing.

Martin-Iverson, M. T., Ortmann, R., & Fibiger, H. C. (1985). Place preference conditioning with methylphenidate and nomifensine. Brain Research, 332, 59-67.

Meyer, R. E., & Mirin, S. M. (Eds.). (1979). The heroin stimulus: Implications for a theory of addiction. New York: Plenum Medical Book Company.

Mucha, R. F., Millan, J. J., & Herz, A. (1985). Aversive properties of naloxone in non-dependent (naive) rats may involve blockade of central beta-endorphin. Psychopharmacology, 86, 281-285.

Mucha, R. F., van der Kooy, D., O’Shaughnessy, M., & Bucenieks, P. (1982). Drug reinforcement studied by the use of place conditioning in rat. Brain Research, 243, 91-105.

Phillips, A. G., & LePiane, F. G. (1980). Reinforcing effects of morphine microinjection into the ventral tegmental area. Pharmacology Biochemistry & Behavior, 12, 965-968.

Phillips, A. G., & LePiane, F. G. (1982). Reward produced by microinjection of (D-ala ), Met  -enkephalinamide into the ventral tegmental area. Behavioural Brain Research, 5, 225-229.

Phillips, A. G., Spyraki, C., & Fibiger, H. C. (1982). Conditioned place preference with amphetamine and opiates as reward stimuli: Attenuation by haloperidol. In B. G. Hoebel & D. Novin (Eds.), The neural basis of feeding and reward. Brunswick, ME: Haer Institute.

Reid, L. D., Hunter, G. A., Beaman, C. M., & Hubbell, C. L. (1985). Toward understanding ethanol’s capacity to be reinforcing: A conditioned place preference following injections of ethanol. Pharmacology Biochemistry & Behavior, 22, 483-487.

Rossi, N. A., & Reid, L. D. (1976). Affective states associated with morphine injections. Physiological Psychology, 4, 269-274.

Schenk, S., Ellison, F., Hunt, T., & Amit, Z. (1985). An examination of heroin conditioning in preferred and nonpreferred environments and in differentially housed mature and immature rats. Pharmacology Biochemistry & Behavior, 22, 215-220.

Schwartz, A. S., & Marchok, P. L. (1974). Depression of morphine-seeking behaviour by dopamine inhibition. Nature, 248, 257-258.

Sherman, J. E., Pickman, C., Rice, A., Liebeskind, J. C., & Holman, E. W. (1980a). Rewarding and aversive effects of morphine: Temporal and pharmacological properties. Pharmacology Biochemistry & Behavior, 13, 501-505.

Sherman, J. E., Roberts, T., Roskam, S. E., & Holman, E. W. (1980b). Temporal properties of the rewarding and aversive effects of amphetamine in rats. Pharmacology Biochemistry & Behavior, 13, 597-599.

Smith, B. R., Amit, Z., & Splawinsky, J. (1984). Conditioned place preference induced by intraventricular infusions of acetaldehyde. Alcohol, 1, 193-195.

Spragg, S. D. S. (1940). Morphine addiction in chimpanzees. Comparative Psychology Monographs, 15, 1-132.

Spyraki, C., Fibiger, H. C., & Phillips, A. G. (1982a). Dopaminergic substrates of amphetamine-induced place preference conditioning. Brain Research, 253, 185-193.

Spyraki, C., Fibiger, H. C., & Phillips, A. G. (1982b). Cocaine-induced place preference conditioning: Lack of effects of neuroleptics and 6-hydroxydopamine lesions. Brain Research, 253, 195-203.

Spyraki, C., Kazandjian, A., & Varonos, D. (1985). Diazepam-induced place preference conditioning: Appetitive and antiaversive properties. Psychopharmacology, 87, 225-232.

Stapleton, J. M., Lind, M. D., Merriman, V. J., Bozarth, M. A., & Reid, L. D. (1979). Affective consequences and subsequent effects on morphine self-administration of d-ala²-methionine enkephalin. Physiological Psychology, 7, 146-152.

Stewart, R. B., & Grupp, L. H. (1981). An investigation of the interaction between the reinforcing properties of food and ethanol using the place preference paradigm. Progress in Neuropsychopharmacology, 5, 609-613.

Strickrod, G., Kimble, D. P., & Smotherman, W. P. (1982). Met-enkephalin effects on associations formed in utero. Peptides, 3, 881-883.

van der Kooy, D., Kalant, H., Mucha, R. F., & O’Shaughnessy, M. (1983). Motivational properties of ethanol in naive rats as studied by place conditioning. Pharmacology Biochemistry & Behavior, 19, 441-445.

van der Kooy, D., Mucha, R. F., O’Shaughnessy, M., & Bucenieks, P. (1982). Reinforcing effects of brain microinjections of morphine revealed by conditioned place preference. Brain Research, 243, 107-117.

Winer, B. (1971). Statistical principles in experimental design. New York: McGraw-Hill.