In a two-lever, food-rewarded drug discrimination paradigm, behavior seems to be governed by a win-stay/lose-shift rule; rats continue to press the lever that yields food, and, when not rewarded, they shift responding to the alternative lever. Here, we report on the effects that antipsychotics and further neuropharmacological agents exert in those conditions. At higher doses, antipsychotics disrupt most or all behavioral parameters in this paradigm. However, at lower doses, rats may select the appropriate lever with normal latency and accuracy, obtain a first food pellet (i.e., "win"), and then, remarkably, shift responding to the alternative lever ("win-shift"). This suggests that antipsychotics can block the effects of reward selectively, i.e., at doses where the initial, secondarily reinforced behavior including the initiation of lever pressing, remains intact. Indeed, saline-treated rats that are given no reward (i.e., "lose") after having selected a lever, also press the alternative lever ("loseshift"). The induction of selective win-shift is specific to antipsychotics, but it varies greatly among them. Perhaps relating to its alleged "incisive" action on delirium and hallucinations, and, surprisingly, in view of its extrapyramidal actions, acutely administered haloperidol (0.04 -0.08 mg/kg) demonstrates win-shift to an exceptional extent, shared only with the newly proposed agent (3-cyclopent-1-enyl-benzyl)-[2-(2,2-dimethyl-2,3-dihydro-benzofuran-7-yloxy)-ethyl]-amine fumarate (F 15063; 0.31-0.63 mg/ kg); the more sedative antipsychotic chlorpromazine demonstrated little selectivity. The paradigm offers a novel tool to characterize antipsychotics with regard to presumably pathogenic motivational processes; mixed D 2 -antagonist/5-hydroxytryptamine 1A -agonist agents such as F 15063 may conceivably share the powerful antipsychotic action of haloperidol while avoiding the sensitization that develops to extrapyramidal effects of haloperidol and consequent negative symptoms.In a typical drug discrimination (DD) experiment, subjects are trained to discriminate a given training drug from saline. For example, rats can be trained to press one of two levers (drug-appropriate lever; DL) for food in sessions preceded by the injection of the opioid fentanyl and to press the alternative lever (saline-appropriate lever; SL) in sessions preceded by saline injection (Colpaert and Janssen, 1984).Rats that have been trained to discriminate fentanyl from saline can demonstrate an exquisite discrimination; the lever selection that is to be made early in the session is correct in nearly 100% of the sessions and occurs with great speed (short latency) and accuracy [the animal pressing the inappropriate lever little if at all before completing the first fixed ration (FR)10 schedule on the appropriate lever]. After having selected the appropriate lever, the rats continue to lever press, totaling some 1500 responses/session, which typically are all made on the appropriate, reward-associated lever (Colpaert, 1978(Colpaert, ...