Drug craving, the desire to experience the effect(s) of a previously experienced psychoactive substance, has been hypothesized to contribute significantly to continued drug use and relapse after a period of abstinence in humans. In more theoretical terms, drug craving can be conceptualized within the framework of incentive motivational theories of behavior and be defined as the incentive motivation to self-administer a psychoactive substance. The incentive-motivational value of drugs is hypothesized to be determined by a continuous interaction between the hedonic rewarding properties of drugs (incentive) and the motivational state of the organism (organismic state). In drug-dependent individuals, the incentive-motivational value of drugs (i.e., drug craving) is greater compared to non-drug-dependent individuals due to the motivational state (i.e., withdrawal) developed with repeated drug administration. In this conceptual framework, animal models of drug craving would reflect two aspects of the incentive motivation to self-administer a psychoactive substance. One aspect would be the unconditioned incentive (reinforcing) value of the drug itself. The other aspect would be relatively independent of the direct (unconditioned) incentive value of the drug itself and could be reflected in the ability of previously neutral stimuli to acquire conditioned incentive properties that could elicit drug-seeking and drug-taking behavior. Animal models of drug craving that permit the investigation of the behavioral and neurobiological components of these two aspects of drug craving are reviewed and evaluated. The models reviewed are the progressive ratio, choice, extinction, conditioned reinforcement and second-order schedule paradigms. These animal models are evaluated according to two criteria that are established herein as necessary and sufficient criteria for the evaluation of animal models of human psychopathology: reliability and predictive validity. The development of animal models of drug craving will have heuristic value and allow a systematic investigation of the neurobiological mechanisms of craving.

The reinforcing properties of cocaine are probably mediated by the mesocorticolimbic dopamine pathways in the central nervous system, but not all of the dopamine receptor subtypes involved in cocaine's reinforcing actions have been clearly identified. Recently, the D-3 receptor has been cloned, and its distribution in the brain has been found to be relatively restricted to limbic projections of the midbrain dopamine system. The D-3-selective compounds 7-hydroxy-N,N-di-n-propyl-2-aminotetralin (7-OHDPAT) and quinpirole potently decreased cocaine self-administration in the rat at doses that were not by themselves reinforcing. Moreover, three dopamine receptor agonists had affinities for binding to the D-3 receptor that correlated highly with their relative potencies in decreasing cocaine self-administration. The D-3 receptor may be involved in the reinforcing effects of cocaine and may be a useful target for the development of new pharmacotherapies for cocaine abuse.

The startle reflex is a contraction of the skeletal and facial musculature in response to an intense sensory stimulus. While the 'primary' neural control of startle involves brain structures at, or below, the level of the mesencephalon, the startle reflex (SR) exhibits several forms of plasticity that are modulated by the forebrain. Sensorimotor gating of the SR occurs when the reflex is inhibited by a weak 'pre-pulse' that occurs 30-500 ms prior to the startling stimulus. Since 'pre-pulse inhibition' (PPI) of startle may be impaired in certain psychiatric and neurologic disorders (e.g. schizophrenia, schizotypal personality disorder and Huntington's disease), there has been considerable interest in determining the neural substrates of this form of startle plasticity. In rats, PPI is modulated by neural elements linking the limbic cortex with the striatum and pallidum. These substrates may include hippocampal glutamate efferents to the ventral striatum and striatal GABAergic efferents to the ventral pallidum. The striatal dopaminergic modulation of PPI appears to involve primarily D2, but not D1, receptors. Pallidal efferents may impinge directly on the 'primary' startle circuitry via projections to the mesencephalon or, indirectly, via projections to the thalamus. Evidence is reviewed for other neurochemical substrates of PPI-including acetylcholine and opiates. Sensorimotor gating of the startle reflex appears to have a discrete and identifiable set of neural substrates that may be important for our understanding of neuropsychiatric disorders characterized by deficient suppression or 'gating' of sensory, cognitive or motor processes.

Dopamine receptor subtypes have been classified generally as D1-like (e.g., D1, D5) or D2-like (D2, D3, D4), and converging evidence suggests that D2-like receptors may be especially important in mediating the abuse-related effects of cocaine. However, it has been difficult to differentiate the roles of the D2-like receptor subtypes in the behavioral effects of cocaine because of the relatively low selectivity of drugs for D2, D3, and D4 receptors in vivo. The goal of the present series of studies was to investigate the contributions of D2-like receptor subtypes in the reinforcing effects of cocaine using new genetic and pharmacological tools. First, we evaluated cocaine self-administration behavior, and related effects of nonselective D2-like drugs, in mutant mice that lack the D2 receptor but express D3 and D4 receptors. When high doses of cocaine on the descending limb of the cocaine dose-effect function were available, D2 mutant mice self-administered at higher rates than their heterozygous or wild-type littermates, but the ascending limb of the cocaine dose-effect function did not differ between genotypes. Elevated rates of drug intake were not attributable to nonspecific increases in response rate, because response rates maintained by presentation of a range of food concentrations were significantly lower in D2 mutant mice than in wild-type mice. In wild-type mice, pretreatment with the D2-like antagonist eticlopride increased rates of self-administration of high doses of cocaine, and the D2-like agonist quinelorane served as a positive reinforcer when substituted for cocaine. However, these effects of eticlopride and quinelorane were not observed in mice that lacked the D2 receptor. Next, we compared the effects of novel antagonists selective for different D2 receptor subtypes on cocaine self-administration behavior in outbred rats. In rats, a D2 selective antagonist increased rates of self-administration of high doses of cocaine and also combinations of cocaine and the D2-like agonist quinelorane, whereas D3/D4 antagonists were ineffective. Collectively, these findings suggest that the D2 receptor is not necessary for cocaine self-administration, but this receptor subtype is involved in mechanisms that limit rates of high-dose cocaine self-administration. Our results also suggest that D3 and D4 receptors do not play major roles in the modulation of cocaine self-administration by D2-like drugs.