There is a good deal of clinical evidence suggesting that compulsion to resume drug taking is an important part of the addiction syndrome. The symptoms comprising motivation to resume drug use, namely craving and compulsion, have been studied experimentally in human subjects. While much work remains to be done, there is evidence showing that these symptoms are influenced by learning. The research has been guided by animal studies demonstrating that drug effects can be conditioned. Much attention has been directed toward demonstrating the existence of drug conditioning in human addicts and exploring the neurological structures that may underlie such learned responses. We do not yet know the relative importance of learning in the overall phenomenon of relapse, and treatments based on conditioning principles are still under investigation.

Repetitive use of psychoactive drugs produces a variety of learned behaviors. These can be classified in the laboratory according to an operant/classical paradigm, but in vivo the two types of learning overlap. The classically conditioned responses produced by drugs are complex and bi-directional. There has been progress in classifying and predicting the types of conditioned responses, but little is known of mechanisms. New techniques for understanding brain function such as micro-dialysis probes in animals and advanced imaging techniques (PET and SPECT) in human subjects may be utilized in conditioning paradigms to "open the black box." Because the existence of conditioned responses in drug users is now well established, clinical studies have been instituted to determine whether modification of conditioned responses can influence clinical outcome. A recently completed study in cocaine addicts has produced evidence that outcome can be improved by a passive extinction technique over an 8-week outpatient treatment program.

Subjects with a history of free-basing and smoking cocaine but no history of opiate injections were exposed to three sets of stimuli. They received cocaine-related stimuli in one session, opiate-related stimuli in a second session, and non-drug stimuli on a third occasion. Compared to the opiate and non-drug cues, the cocaine-related events caused reliable decreases in skin temperature and skin resistance, and reliable increases in heart rate, self-reported cocaine craving, and self-reported cocaine withdrawal. Furthermore, control subjects lacking a history of cocaine or opiate use failed to show such differential responding. These results suggest that cocaine-related stimuli evoke Pavlovian conditioned responses in cocaine abuse patients. Such findings encourage continuing efforts to develop drug treatment strategies based on conditioning principles.