Classically Conditioned Phenomena in Human Opiate Addiction


Classical and operant conditioning factors are both potentially significant in the maintenance of opiate use. Analysis from the perspective of the operant conditioning paradigm emphasizes the importance of discriminative stimulus control and the efficacy of opiates as reinforcers (). In the context of the classical conditioning paradigm, emphasis is placed on environmental correlates of drug effects and withdrawal symptoms as elicitors of overt behavioral and physiological responses. Concurrently it must be recognized that a model based on integration of both paradigms probably reflects most accurately the reality of human opiate dependence ().

In the context of either the operant or classical conditioning paradigms, seemingly contradictory and diverse effects of stimuli and events may be identified. However, careful analysis leads to the conclusion that systematic results prevail and that findings parallel those involving other behaviors and reinforcers. As has been discussed in a recent review (), the primary problems appear to arise in delineating the phase of opiate action (e.g., onset, termination, withdrawal) with which stimulus events are associated. A secondary problem arises in differentiating patterns of use and determining presence or absence of dependence and the withdrawal syndrome. That is, certain behavioral features, development of conditioned correlates, and hence the nature of explanatory concepts are related to certain aspects of drug effects or sequelae.

Two major categories of events are associated with chronic administration of opiates. One is drug onset, with its diverse physiological effects. The second is characterized by the myriad physiological and behavioral responses in the physically dependent organism following termination of a regimen of opiate administration.

The first major category was the object of early investigations of conditioning phenomena in relation to drug effects. These involved the classical conditioning paradigm and focused on conditioned responses correlated with the drug effects. A neutral stimulus was presented in temporal contiguity with an injection of morphine. After repeated pairings the formerly neutral stimulus became clearly established as a conditioned stimulus (CS) as indicated by physiological or behavioral changes elicited when it was presented without the morphine injection. This response or group of responses, termed the conditioned response (CR), typically reflected, but was not necessarily identical to, the unconditioned effects of morphine. The seminal work of Pavlov entailed analysis of this aspect of conditioning in relation to drug effects. Much of the operant self-administration research likewise involves examination of aspects of conditioning in relation to drug onset effects.

The second major category of experiments entails analysis of two aspects of phenomena generally associated with physical dependence. One class of experiments involves analysis of conditioned stimuli established in relation to the opiate withdrawal syndrome which reliably emerges at some point following administration of the last opiate dose in the physically dependent Subject. In these experiments the environmental stimulus previously paired with either antagonist- or metabolism-induced withdrawal symptoms elicits similar symptoms. That is, the conditioned response resembles the unconditioned responses of the abstinence syndrome. In a second class of experiments for which the observed response is similar to withdrawal syndrome, the phenomenon under study is “conditioned tolerance.” The experimental design is similar to that in which conditioned drug effects are examined. The resultant CR, however, is abstinence-like and is termed a “counteradaptive” () or “conditioned tolerance” response (). Although this second class of withdrawal phenomena remains controversial, it continues to be of considerable scientific interest (). While these phenomena have been demonstrated in the laboratory to be robust, the extent to which they obtain in the natural environment of opiate-using patients is unclear; further, the extent to which they may contribute to persistence of opiate use is yet to be objectively determined.

A series of investigations directed at the analysis of conditioned responses in opiate-dependent or postdependent human sujects has been implemented at our University of Pennsylvania/Veterans Administration Medical Center laboratory. Analysis has involved responses which have been established either naturally or experimentally. That is, in some cases the circumstances for development of classically conditioned responses have been established in the laboratory, and exposure to these situations has elicited appropriate responses. In other cases the investigations have involved analysis of the behaviors established in the Subjects’ natural environment which are elicited through various experimental manipulations and evaluated in the laboratory setting. Evidence has emerged for both conditioned drug effects and conditioned withdrawal effects. The focus of the current discussion will be elicitation of withdrawal-like responses and description of the circumstances under which these responses evolve.

Opiate-like conditioned responses

There have been several clinical reports of what may be identified as “conditioned drug responses” associated with the behavior of self-injection (). This so-called “needle-freak” phenomenon, which also represents a clear “placebo response,” is characterized by pleasurable subjective effects during and following self-administration of saline or other pharmacologically inert solutions. Presumably conditioned physiological changes also occur, and the analysis of these responses is of considerable interest since it clearly indicates the potential importance of drug-use-correlated behaviors serving as conditioned reinforcers. Although this phenomenon has not been studied systematically, it has been observed that some subjects exhibited morphinelike subjective and physiological effects when saline was self-injected (). Typically, these opiate-like effects follow self-injection rather than occurring during the pre-injection period. Meyer and Mirin () have also reported opiate-like post-injection autonomic changes in 11 of 22 patients who self-injected heroin while opiate agonist effects were blocked by naltrexone. As shown in figure 1, the injection ritual (CS1) may act as a complex conditioned stimulus which, after repeated pairing with opiate agonistic effects (UCSl), attains the ability to evoke weak agonistic effects (CRl).

The foregoing reports and observations parallel studies with animals in which environmental stimuli previously associated with drug effects can serve as conditioned reinforcers or classically conditioned elicitors of behavior. Thus, for example, Woods and Schuster () noted that a stimulus previously associated with drug self-administration in monkeys can, for a period of time, maintain responding for saline infusions. Similarly, but in the context of the classical conditioning paradigm, La1 and coworkers () reported that a stimulus previously paired with drug effects can attenuate a component of the opiate withdrawal syndrome. Therefore, it is clear that mechanisms associated with either operant or classical conditioning may directly or indirectly contribute to persistence in responses associated with drug-correlated stimuli. In turn, it can be argued that conditioned drug effects may contribute to the general pattern of persistence observed in human drug-seeking behavior.

Opiate withdrawal-like conditioned responses

Since one phase of opiate action (i.e., onset) can serve as the basis for a conditioned response and can generate behaviors associated with opiate self-administration, conditioned responses may also be expected to emerge in relation to the other major physiological/behavioral event associated with opiate use–that is, the withdrawal syndrome. Wikler’s early observations led to the proposal that conditioned responses might serve to generate drug-seeking behavior. Thus, for example, conditioned withdrawal has been presumed to underlie case reports involving drug-free postdependent patients who exhibit physical and subjective evidence of opiate withdrawal when they return to the environment in which drugs were used (). In the laboratory the conditioned withdrawal phenomenon has been established in patient volunteers maintained on a methadone regimen by pairing naloxone-precipitated withdrawal (unconditioned response, UCR) with a novel stimulus (CS). The resultant CR resembles the UCR (opiate withdrawal) (). Similar physiologic, behavioral, and subjective responses have been observed when methadone-maintained patients were exposed to videotaped sequences of themselves using drugs in the laboratory () and when stimuli such as drug-related videotapes, slides, or objects were shown to drug-free or methadone-maintained patients (). Sideroff and Jarvik () also reported that both physiologic and subjective withdrawal effects occur in patients undergoing detoxification after viewing a videotape of drug use by other individuals.

Analysis of the above-described phenomena is of course difficult and complex, and unresolved questions exist despite numerous and repeated observations of dependent and postdependent patients in the laboratory setting. For example, it appears that typically when patient SubJects perform the pre-injection rituals, i.e., “cook up” (drug preparation) and “tie off” (tourniquet application), their physiologic, subjective, and behavioral responses resemble opiate withdrawal (). Only rarely have drug-like responses been observed. Several factors may contribute to the observed preponderance of “withdrawal-like” responses. First, it should be noted that the physiological concomitants of opiate withdrawal are similar to those of nonspecific autonomic arousal reactions evidenced when individuals with no drug use history view drug-related stimuli, including videotapes of individuals injecting drugs.

Second, it is at times difficult to identify separately the physiological correlates of drug-like and withdrawal-like phenomena since some of the physiological responses (but presumably not mechanisms) are similar. One direct effect of opiates is nausea; indeed, this is sometimes used as an indicant of “street drug” quality and is termed a “good sick” by opiate users (). Yet, nausea is also one of the clearly observable responses during opiate withdrawal. This is just one example of possible problems in analysis. It should be noted that the physiological response changes correlated with stimulus presentations for patients tend to be more durable and robust than in drug-naive subjects.

A third consideration is the possibility that responses elicited by pre-injection stimuli differ from those immediately post injection. That is, withdrawal-like responses may precede injection, while conditioned drug-like responses may follow placebo self-administration. This issue too requires further analysis, and the data of Eikelboom and Stewart () suggest that more refined analyses may permit differentiation of these phenomena.

It was noted that withdrawal-like responses may emerge via a second mechanism reflected by “conditioned tolerance” (Siegel 1976, 1978), although this explanation is at present tenuous. Figure 1 shows conditioning paradigms which could be expected to produce withdrawal-like CR’s in dependent subjects. One procedure involves opiate withdrawal occurring as the opiate is metabolized and opiate receptors are evacuated (UCS2). Since UCS2 is directly paired with drug procurement or pre-injection stimuli (CS2), the pre-injection stimuli acquire the ability to evoke conditioned withdrawal (CR2).

The other procedure leading to withdrawal-like responses involves an adaptive or homeostatic response (UCRlA) to the occupation of the opiate receptors. This adaptive response may be partially responsible for tolerance phenomena because it tends to oppose or diminish the agonistic action of the opiate. The tolerance response UCRlA. reliably follows UCS1 (receptor occupation), so that an adaptive conditioned response (CRlA) may also be conditioned to pre-injection stimuli (CSl).

Since the drug procurement (CS2) and pre-injection rituals (CSl) are usually paired with both receptor evacuation (the UCS for the withdrawal UCR) and subsequently with receptor occupation (the UCS for the compensatory UCR), these stimuli may elicit either adaptive reactions or the withdrawal reactions as conditioned responses. However, in terms of physiological variables, both types of conditioned response resemble opiate withdrawal (UCR2). Thus, what appear as withdrawal responses in a drug procurement area actually may be the result of two different conditioning processes, conditioned tolerance and conditioned withdrawal. Because both may be operative, a redundancy occurs which may increase the apparent strength of the conditioned response observed.

Extinction of conditioned opiate effects in humans

If, as has been suggested, drug-like and drug withdrawal-like effects can be established via operant and classical conditioning, it should be expected that they can similarly be extinguished. That is, if the unconditioned stimulus eliciting classically conditioned responses is no longer paired with the conditioned stimulus or if the operant behavior is no longer reinforced, a decrease in response strength should be expected. Goldbero and Schuster () reported that conditioned withdrawal in morphine-dependent monkeys was quite resistant to extinction. With humans, the laboratory investigations have produced variable results in resistance to extinction. Thus, for example, in the analysis of experimentally conditioned withdrawal in humans using a novel conditioned stimulus and a small number of training trials. O’Brien and coworkers () reported that conditioned-withdrawal symptoms diminished rapidly during repeated unreinforced (test) trials. In contrast to these studies of laboratory conditioned withdrawal responses, studies of withdrawal CR’s in response to naturalistic conditioned stimuli (“cook-up” and “shoot-up” rituals) indicate great resistance to extinction. Presumably, these naturally conditioned CR’s are established during an extensive history of opiate use involving a large number of “training” trials in the user’s natural environment. In addition, a schedule of intermittent reinforcement in the natural environment may evolve with respect to some features of operant behaviors, thereby adding to the complexity of factors contributing to behavioral persistence. These differences between the laboratory and natural environment may contribute to the apparent disparity of resistance to extinction ().

In a double-blind experimental design, the effects of systematic repetition of drug-associated rituals under circumstances in which opiate reinforcement was either absent or blocked by an antagonist have been further examined. Opiate-free post-dependent subjects were permitted, in the laboratory setting, to engage in the pre-injection and self-injection rituals (using opiate or saline) while being maintained on the opiate antagonist naltrexone (). Typically, withdrawal-like autonomic responses were observed during the pre-injection ritual. Initially opiate-like subjective and physiological effects occurred after injection regardless of the contents of the syringe. Subsequently, the opiate-like subjective effects disappeared (extinguished) after a few unreinforced trials. Interestingly, on later trials, when opiate-like effects no longer occurred, the injection ritual was followed by an increase in the autonomic withdrawal-like responses ().

In another study () it was found that detoxified patients who were allowed to self-inject either opiate or saline showed compensatory (i.e., withdrawal-like) autonomic changes prior to self-injection and opiate-like autonomic changes after the injection. In contrast, the same subjects given an unsignalled infusion of an opiate showed only the opiate-like changes and these effects only occurred after the infusion. These findings are strikingly similar to those of Eikelboom and Stewart (), who reported differing response patterns to the two stimulus conditions previously paired with pre-injection withdrawal responses and post-injection drug effects. Thus in both human subjects (postdependent patients) and a nonhuman species, the pre-injection stimuli elicited withdrawal-like responses and the post-injection stimuli elicited opiate-like changes. Whether or not these differential responses reflect conditioned tolerance or conditioned compensatory responses prior to injection is, of course, unclear. Both patterns of responses can be observed in experimental paradigms designed to reflect a sequence of events which prevails in the natural environment for some opiate users. Efforts to eliminate both patterns of responding using extinction procedures in the laboratory thus far suggest that the conditioned positive reinforcing effects diminish more rapidly than the conditioned withdrawal responses. In addition it appears that elements of both response patterns may persist in some human subjects (). The difficulties in experimental analysis of these phenomena are considerable, and it is clear that further research is required.


Laboratory investigations with postdependent and dependent human Subjects, as well as studies with rodents and primates, have demonstrated that behavioral and physiological correlates of drug and withdrawal responses can be conditioned using operant and/or classical procedures. In addition it appears that the behaviors and physiological responses of the several phenomena may coexist and be evident concurrently or sequentially in the same Subject. Presumably when the more complex patterns emerge, as is typical with human Subjects who are former opiate users, the results reflect the complexities of an extensive and variable past history. Many factors potentially may influence the relative strengths of both operant and respondent behaviors. For example, opiate users who are not physically dependent because they use insufficiently high daily drug doses may experience unconditioned withdrawal rarely or not at all. These subjects could be expected to show mainly opiate-like CR’s; theoretically, however, some compensatory CR’s should also occur, albeit less intensely. On the other hand, for subjects who are physically dependent, two mechanisms for withdrawal-like conditioned responses () exist: compensatory CR’s and withdrawal CR’s. This redundancy may explain why, in investigations with some human opiate users, withdrawal-like CR’s dominate and are difficult to extinguish.

While it is apparent that conditioning in opiate users is complex and incompletely understood, the effects appear strong enough to have clinical implications. Studies involving opiate-using patients or postdependent individuals present special problems because of their variable “training” histories. Perhaps further research with both humans and animals will elucidate the conditions under which the different types of condioned responses are most likely to develop.


Selections from the book: “Behavioral Pharmacology of Human Drug Dependence”. Travis Thompson, Ph.D., and Chris E. Johanson, Ph.D., eds. Presents a growing body of data, systematically derived, on the behavioral mechanisms involved in use and abuse of drugs. National Institute on Drug Abuse Research Monograph 37, July 1981.