Designing studies of drug conditioning in humans.

There has been much recent interest in the possibility that signals for drug use in humans acquire the ability to evoke classically conditioned (learned) states which motivate drug taking. Much data now suggest that cues paired with drug use come to elicit physiological responses and subjective reports of drug-related feelings like craving and withdrawal. However, the designs employed do not permit the conclusion that the observed responding results from classical conditioning. Studies which look directly at conditioning in the laboratory by pairing neutral stimuli with drug administration have not provided appropriate controls for unlearned effects such as sensitization or pseudo-conditioning. Similarly, studies which assess responding to cues thought to signal drug use in the natural environment (e.g., the sight of someone injecting heroin) have not adequately assessed whether such cues have unconditioned (unlearned) effects. Determining whether responding to drug-related cues results from classical conditioning has important implications for the development of drug treatments. Consequently, the purpose of the present review is to outline a set of criteria for determining that responses to drug-related stimuli in humans are learned. Existing studies are reviewed in light of these criteria and paradigms for further work are suggested.