available for a fixed amount of time or throughout the day. Drugs that are reinforcing prompt the subject to work more or pay a higher cost for them than for the control; reinforcing drugs also lead to a greater persistent responding for them even when they are no longer available (Henningfield et al., 1991). Typically the dose-response curve is U-shaped (Risner and Goldberg, 1983; Rose and Corrigall, 1997) with low and high doses resulting in reduced drug self-administration. Low doses may produce limited or undetectable effects and high doses may produce adverse effects.
Drug discrimination models involve training the subject to discriminate the stimulus properties of drug A from drug B. A third drug may be introduced, and the animal or human subject is asked to decide whether the drug is more like drug A or drug B (Bigelow and Preston, 1989; Preston, 1991). Subjects can also be trained to discriminate among several sets of drugs or different doses of a drug. This model allows determination of the mechanism of action of a drug. For example, if one wanted to determine whether an opioid has µ agonist or κ agonist activity, an experiment can be developed in which the subject is trained to discriminate between drugs that are known to have each of these activities. After this period of training, the drug in question is introduced and the subject has to indicate whether the drug is more like drug A (e.g., a µ agonist) or drug B (a κ agonist). This model can be also used to determine whether a drug has the stimulus properties of a particular pharmacological class of drugs that are abused. A similar method is used in a drug preference procedure, in which subjects are exposed to drug A and drug B, and are required to self-administer each of these drugs during separate experimental sessions. After the drug exposure or sampling period, subjects are then asked to choose between drugs A and B, to determine their preference for one drug or another (de Wit, 1991). Drug A or B can be two different doses of a drug, different types of drugs, or an active and placebo drug.
The conditioned place preference model also is used to determine the abuse liability of a drug. Animals are trained that the drug is available only in a particular place (e.g., a specific chamber). Then a determination is made of how frequently the animal is willing to go to this place. If it is chosen significantly more frequently than the other place which is associated with a control drug or no drug administration, the experimental drug may have abuse potential (Bozarth, 1987).
Drug withdrawal models have typically involved observing signs and symptoms during a period of abstinence after repeated admin-