The major contribution of behavioral research to the study of drug abuse has been the development of the self-administration model and the use of this model to test for abuse liability and to expand our understanding of addiction. This basic model has been augmented by other models based on the principles of learning and conditioning such as drug classification (drug discrimination); the relationship between drug use and variables controlling use (behavioral economics); the nature of transition states in drug abuse (initiation, abstinence, withdrawal); motivational states (e.g., incentive motivation); and the roles of tolerance and physical dependence in drug-seeking behavior.

The drug self-administration model is based on the learning principle that behavior is maintained by its consequences, called reinforcers. Laboratory animals (humans and nonhumans) will work to receive a range of different drugs administered orally, intramuscularly, intravenously, by smoking, or by insufflation. In this model, the laboratory animal performs some action, such as depressing a lever, to trigger the administration of a drug (e.g., through an indwelling catheter or a solution to drink). In general, those drugs (e.g., cocaine, heroin, nicotine, alcohol) that maintain drug taking in nonhumans are also commonly abused by humans, and those that are avoided by humans (e.g., antipsychotics) are also avoided by nonhumans. These results are replicable in virtually every species tested with the model and with different routes of administration. Such findings brought into question the traditional explanations of the etiology of drug abuse, such as psychopathology or various social deprivations.

This model also allows behavioral researchers to control past history and current environmental conditions, thus demonstrating that it is the interaction of the drug's pharmacological effects with past history and current environmental conditions (i.e., setting) that determines whether sampling an abusable drug will proceed to persistent use or abuse (e.g., Barrett and Witkin, 1986). This model points to the importance of a confluence of variables in drug-taking behavior and has broadened the clinician's understanding of the various causal factors that might be involved in drug abuse.

The drug discrimination paradigm is considered a model of the subjective effects of drugs in humans. In this paradigm, research subjects are