New Perspectives on Cocaine Addiction: Findings from Animal Research Michael A. Bozarth Department of Psychology State University of New York at Buffalo Buffalo, New York 14260 Research with laboratory animals has provided several insights into the nature of cocaine abuse and addiction. First, the nature of drug addiction has been re-evaluated, and the emphasis has shifted from physical dependence to compulsive drug-taking behavior. Second, animal studies suggest that cocaine is at least as addictive as heroin and possibly even more addictive. Third, cocaine is potentially more dangerous than heroin as evidenced by the higher fatality rate seen in laboratory animals given unlimited access to these drugs. Fourth, the neural basis of cocaine reinforcement has been identified and involves an enhancement of dopaminergic neurotransmission in the ventral tegmental dopamine system. Other addictive drugs (e.g., opiates) may also derive at least part of their reinforcing impact by pharmacologically activating this reward system. Fifth, although the biological consequences of repeated cocaine self-administration on central nervous system functioning are poorly understood, preliminary findings suggest that intravenous cocaine self-administration may decrease neural functioning in this brain reward system. This has important clinical implications because diminished functioning of an important brain reward system may significantly contribute to relapse into cocaine addiction. These and other findings from experimentation with laboratory animals suggest new considerations for the etiology and treatment of drug addiction. Drug addiction is a major health problem in North America affecting increasingly larger segments of society. The incidence of cocaine usage has dramatically increased over the past decade (e.g., Adams and Durell, 1984; Kozel and Adams, 1986), and heroin addiction, once thought to be an affliction of lower socio-economic groups, has extended to a much broader segment of society. The Division of Epidemiology and Statistical Analysis of the National Institute on Drug Abuse has aptly documented the current trends in cocaine use and provides the most comprehensive evaluation of the current problem in the United States. An estimated 5.8 million people have used cocaine in the past 30 days, and the lifetime prevalence of cocaine use is estimated to be 22.2 million (Kozel and Adams, 1986). Emergency room admissions for cocaine-related problems has risen 800% in the past decade (Schuster, 1988), and the cost of cocaine trafficking is estimated to be between $50 and $70 billion annually (Clayton, 1985). The current number of people seeking treatment for cocaine addiction far exceeds the available treatment facilities in the United States. In addition, there are pronounced cyclic variations in the abuse patterns of other drugs (e.g., amphetamine, hallucinogens, phencyclidine), although the epidemic growth of cocaine abuse is currently the most significant problem. With the introduction of readily available freebase cocaine ("crack"), the dramatic increases in cocaine abuse are likely to continue. Drug Addiction The term drug addiction describes a situation where the use of a drug has escalated to a level where drug procurement and use seem to dominate the individual’s behavior. Addiction is thus defined by the behavioral pattern of drug use (Jaffe, 1975; see also Bozarth, 1987a), and it is characterized by compulsive drug use that is necessary for maintaining optimal psychological functioning of the individual. Attempts to identify the underlying cause of drug addiction have confused the development of physical dependence with the addiction itself. Although it is true that addiction to many compounds (most notably the opiates) is accompanied by the development of physical dependence, this is not a defining characteristic of addiction. Research with laboratory animals and human addiction to drugs, such as cocaine, that do not produce a classic withdrawal syndrome when their use is terminated have prompted a re-evaluation of the role of physical dependence in drug dependence. It is therefore important to distinguish physical dependence from drug addiction qua addiction and to define drug addiction according to its behavioral characteristics (viz., compulsive drug-taking behavior). Animal Models The ability of a drug to control behavior is probably the most striking property of an addiction. Animal models of human drug addiction have focused on this property. Addictive drugs can control behavior in much the same manner as conventional reinforcers, when drug delivery is made contingent upon some operant response, such as lever pressing (Johanson, 1978; Spealman and Goldberg, 1978; see also Fischman and Schuster, 1978). Most drugs that are addictive in humans are readily self-administered by laboratory animals, while drugs that lack addictive properties in humans are usually not self-administered by animals (Deneau, Yanagita, and Seevers, 1969; Griffiths and Balster, 1979; Weeks and Collins, 1987; Yokel, 1987). The illicit use of drugs seems to be related to their reinforcing or rewarding properties (see Bozarth, 1987a; Fischman and Schuster, 1978; Jaffe, 1975), and attempts to develop animal models of drug addiction have focused on drug self-administration methods (e.g., Deneau et al., 1969; Johanson, Woolverton, and Schuster, 1987; Schuster and Johanson, 1974; Woods and Schuster, 1968; see also Bozarth, 1987b; Krasnegor, 1978). Of the several techniques available for studying drug self-administration, intravenous self-administration is probably the most widely accepted method (Bozarth, 1987a, 1987b).Psychomotor stimulants (e.g., amphetamine, cocaine) have strong addictive properties in humans and are readily self-administered by laboratory animals. Together with the opiates (e.g., morphine, heroin), they constitute prototypical agents for the study of drug reinforcement in laboratory animals. Numerous studies have examined the behavioral effects of these drugs using the intravenous self-administration paradigm, and considerable knowledge has been gained regarding their ability to reinforce and control behavior. In the past decade, the neural basis of reinforcement from psychomotor stimulants has been delineated, and this work will be briefly reviewed here. In addition, preliminary evidence suggesting that cocaine self-administration may produce alterations in the brain system mediating its reinforcing effects will also be presented. Addiction Liability The term addiction liability refers to the drug’s tendency to produce an addiction. This is not equivalent to measuring the motivational strength of the drug which might be construed as measuring the strength of the addictive behavior after the addiction has developed. If a drug is capable of generating addictive behavior and if the development of addictive behavior is related to the extent of drug exposure, a critical variable for the genesis of the addiction is likely to be the initial reinforcing action of the drug. One measure of a drug’s addiction liability may be its tendency to establish self-administration behavior. About 80% of the rats tested under standard laboratory conditions (i.e., 6 hours/day, 5 days/week) learn to intravenously self-administer cocaine in the first 20 days of testing; testing for the acquisition of intravenous heroin self-administration behavior reveals a similar percentage of animals learning to take the drug (M. Bozarth and R. Wise, unpublished observations). In these tests the animals were not shaped to lever press and were administered only a single daily priming injection if they failed to initiate self-administration in the first 15 minutes of testing. The percentage of animals learning to self-administer cocaine or heroin probably approaches 100% if operant shaping procedures are used. This observation suggests that cocaine and heroin have a similar addiction liability. If the speed of acquiring drug self-administration behavior is used to compare the relative addiction liability of cocaine and heroin, cocaine appears to be somewhat more addictive than heroin since 70% of the animals learn to self-administer cocaine and only 30% of the animals reliably self-administer heroin by the end of the first 5 days of testing. More extensive testing over a wide range of drug doses and various exposure conditions is necessary to determine the general applicability of this conclusion. A method of determining the strength of drug-taking behavior involves progressive-ratio testing. With this method animals are trained on a partial reinforcement schedule (viz., fixed ratio), and the number of lever presses required for each drug injection is increased until the animals fail to self-administer the drug (see Brady, Griffiths, Hienz, Ator, Lukas, and Lamb, 1987; Yanagita, 1987). This method determines the "break point" or maximum number of responses that the animal will make for a single drug injection. Progressive-ratio testing essentially measures the strength of a behavior and hence, it is an indicator of the motivational properties of the drug. Progressive-ratio testing may appear to provide an alternative method of assessing a drug’s addiction liability, but it is probably not as appropriate as simply measuring the tendency of the compound to establish self-administration behavior. Addiction can be divided into two phases—acquisition and maintenance phases. Progressive-ratio testing determines the strength of the behavior after drug-taking behavior (addiction?) has been established. If the drug is a powerful reinforcer, it is potentially an addictive drug. However, the critical variable in determining the tendency to develop an addiction is probably related to the drug’s initial rewarding properties. Drugs that can eventually become powerful reinforcers (e.g., as measured by the progressive-ratio technique) but that are not initially strong reinforcers (e.g., as measured by the speed of establishing drug self-administration) are less likely to be addictive than compounds that are powerful reinforcers and lead to rapid acquisition of drug-taking behavior. Thus, once it is established that a compound can serve as a strong reinforcer, which is the case for most compounds that are addictive in humans, the relative speed of acquiring drug self-administration behavior is likely to be the best single indicator of the drug’s addiction liability. (Considering the factors of drug availability, social custom, and pervasive media exposure, addiction to ethanol is probably much less widely spread than would be expected if it were initially a strong reinforcer; cf., predicted incidence of cocaine or heroin addiction if similar conditions prevailed.) Using this criterion, intravenously administered cocaine is at least as addictive as heroin and perhaps even more addictive. Reward Mechanisms Behavior controlled by its consequences and the specific relationship between a behavioral response and some stimulus event (i.e., reinforcer) is generally subsumed under the topic of operant conditioning and reinforcement theory. If the frequency of a behavior is increased by the presentation of some stimulus, the stimulus is serving as a positive reinforcer. If the frequency of a behavior is increased by the removal of some stimulus, the stimulus is serving as a negative reinforcer. Both cases of reinforcement involve an increase in the behavioral response that the reinforcer is associated with. Events associated with positive reinforcement are usually appetitive stimuli and are frequently considered pleasant by humans, while events associated with negative reinforcement are usually aversive stimuli and are considered noxious by humans. The case where a behavioral response is followed by the presentation of an aversive stimulus is termed punishment and usually produces a decrease in the behavior paired with the aversive stimulus. Punishment effects are probably not important in the study of drug addiction (except possibly in inhibiting drug intake; but see Mello, 1978) and will not be considered in this paper. Most studies of intravenous self-administration are presumed to involve positive reinforcement mechanisms—the drug delivery is made contingent upon an operant response such as lever pressing, and the animal appears to be working for some appetitive consequence of the drug’s pharmacological action. It is possible, although not documented, that in some cases drug self-administration may be partially controlled by the drug’s ability to relieve an aversive condition produced by repeated drug administration. Because the positive reinforcing characteristics of drug administration are best understood, the neural basis of this mechanism will be considered first. Positive Reinforcement Despite differences in their chemical structures, cocaine and amphetamine have very similar effects on catecholamine (i.e., dopamine and norepinephrine) systems. Cocaine blocks the reuptake of catecholamines (Heikkila, Orlansky, and Cohen, 1975), and amphetamine blocks their reuptake and inhibits their degradation by monoamine oxidase (Axelrod, 1970; Carlsson, 1970). The result of these actions is an enhancement of catecholaminergic neurotransmission. Although other neurotransmitter systems are also affected by cocaine and amphetamine, the enhancement of brain catecholaminergic neurotransmission has been related to several important behavioral actions of these compounds. Early work with psychomotor stimulant self-administration demonstrated that the central dopamine-enhancing effects of amphetamine and cocaine were primarily responsible for their reinforcing effects in animals. Injections of neuroleptic drugs, which block dopamine receptors, attenuate the reinforcing effects of intravenous stimulants in animals (de Wit and Wise, 1977; Yokel and Wise, 1975). In comparison, drugs that block noradrenergic receptors have little effect on intravenous stimulant self-administration (de Wit and Wise, 1977; Yokel and Wise, 1976). Little work has been done in humans, but Gunne (1977) has reported that neuroleptic injections attenuate the euphoria produced by amphetamine, and several investigators have suggested that euphoria in humans closely parallels the reinforcing effects of these compounds in animals (e.g., Griffiths and Balster, 1979; Henningfield, Johnson, and Jasinski, 1987). Thus, even though psychomotor stimulants enhance both dopaminergic and noradrenergic neurotransmission, only the dopamine-enhancing action is critically involved in their reinforcing effects. The neural site of action for psychomotor stimulant reward has also been identified. Dopamine-depleting lesions of the nucleus accumbens (Lyness, Friedle, and Moore, 1979; Roberts, Corcoran, and Fibiger, 1977; Roberts, Koob, Klonoff, and Fibiger, 1980) or of the ventral tegmental area (Roberts and Koob, 1980) disrupt stimulant self-administration in animals, while lesions of the noradrenergic systems are ineffective (Roberts et al., 1977; see also Roberts and Zito, 1987). In addition, amphetamine is self-administered directly into the nucleus accumbens (Hoebel, Monaco, Hernandez, Aulisi, Stanley, and Lenard, 1983). There is some evidence to suggest that the frontal cortex may also be involved in stimulant reward (Goeders and Smith, 1983; Phillips, Mora, and Rolls, 1981), but disagreement exists regarding the importance of this site in stimulant reinforcement. Nonetheless, all three brain regions implicated in stimulant reward are part of the same ventral tegmental dopamine system. This dopamine system has its cell bodies in the ventral tegmental area and sends its axons rostral to terminate in the nucleus accumbens and in the frontal cortex (Lindvall and Björklund, 1974; Ungerstedt, 1971a). Although additional work is necessary to assess the relative importance of the two primary terminal projections of this system, the consensus is that this brain system is critically involved in the reinforcing effects of psychomotor stimulant drugs (see Bozarth, 1985, 1986a). More recent work has also implicated this same dopamine system in the mediation of opiate reward (for reviews, see Bozarth, 1985, 1986a, 1987c; Bozarth and Wise, 1983a; Wise and Bozarth, 1981, 1982, 1984). Figure 1 illustrates the common reward substrate involved in psychomotor stimulant and opiate reinforcement. A recent study has shown a partial cross-substitution of psychomotor stimulant and opiate rewards (Bozarth and Wise, 1986), thus providing an important test of this proposed model. In addition to an important role in drug reward, brain dopamine systems are involved in other behaviors such as locomotor activity (Breese, Hollister, and Cooper, 1976; Costall, Domeney, and Naylor, 1984; Creese and Iversen, 1973, 1975; Ungerstedt, 1978), stereotypy (Creese and Iversen, 1973, 1975) and feeding behavior (Ungerstedt, 1971b, 1978; Wise, 1982). The neural substrates of these effects have been partly identified, and at least the brain system involved in locomotor activity seems to be the same as that involved in stimulant reinforcement (see Wise and Bozarth, 1987). Other Influences In addition to the well documented positive reinforcing effects of cocaine and other addictive drugs, other factors may contribute to the net reinforcing impact of drug self-administration. Two possible factors may involve the same brain reward substrate as that mediating the positive reinforcing effects of cocaine, but repeated drug administration would be necessary for these effects to emerge. Negative Contrast Effect Animals given a fixed magnitude of reward for performing some operant or instrumental response will produce a constant level of behavioral output. If the magnitude of the reward is abruptly shifted, rapid changes in the animal’s performance are produced that are not solely dependent on the magnitude of the reward, but rather, they are related to the degree of change in reinforcement magnitude. For example, animals given a certain level of food reinforcement for running an alley will perform at a constant speed. If the reward magnitude is suddenly shifted to a lower level, the animals’ running speed will be much lower than if they had been receiving the lower level of reinforcement with each trial. This phenomenon is called a negative contrast effect and was originally described by Crespi (1942) for laboratory animals running an alley to receive a food reinforcement. If cocaine super-potently activates a brain reward system involved in the control of other motivated behaviors, then normal, less potent activation of this system by natural rewards may not govern behavior as efficaciously following repeated activation by cocaine. The lower level of activation produced by natural rewards may parallel the abrupt shift to a lower reinforcement magnitude producing a negative contrast effect. Thus, stimuli normally capable of governing behavior may not be very efficacious following repeated cocaine administration. This may explain the amotivational syndrome and diminished interest in other normally rewarding events experienced by chronic cocaine users. The person would then be further motivated to experience the super-potent rewarding effects produced by pharmacological activation of this brain reward system. Possible Negative Reinforcement Another motivational influence that may develop with repeated drug administration is negative reinforcement. This process involves an aversive condition that is terminated by drug administration. In the case of opiates, one possible negative reinforcement mechanism may be related to the development of physical dependence. The repeated administration of high doses of opiates produces physical dependence, and drug abstinence is followed by a well characterized withdrawal syndrome. This withdrawal state is generally presumed to be aversive, and it can be readily terminated by opiate administration. The relief from withdrawal distress can potentially provide negative reinforcement for repeated opiate self-administration. It should be noted that negative reinforcement mechanisms, long presumed by many clinicians to be critical for opiate addiction, have not been shown to contribute to opiate reinforcement. In fact, the neural substrates mediating the rewarding and physical dependence-producing effects of opiates are anatomically dissociable (Bozarth and Wise, 1984). This does not preclude, however, the possibility that negative reinforcement mechanisms can contribute to the net reinforcing effect of drug self-administration (see Bozarth, 1988; Bozarth and Wise, 1983a), but they clearly are not necessary for opiates to be reinforcing (Bozarth and Wise, 1983b, 1984). Unlike opiate withdrawal, strong physiological reactions accompanying cocaine withdrawal have not been reported. Thus relief from cocaine withdrawal, at least as defined by autonomic disturbances, is not likely to be a significant factor in cocaine addiction. Another possibility is that repeated cocaine use may alter neural functioning in the ventral tegmental reward system. Cocaine might produce an impaired level of function necessitating continued cocaine administration to restore normal functioning of this reward substrate. In this way, cocaine could have a normalizing effect that could be considered a negative reinforcement mechanism. Preliminary empirical evidence of a possible neural basis for this negative reinforcement mechanism will be presented in a later section on Physiological Toxicity. Toxicity The toxicity of cocaine has been generally underestimated by clinicians and the popular media (e.g., Seiden, 1985; Strategy Council on Drug Abuse, 1973). The widespread use of this compound makes the study of its toxicology particularly important. Although repeated exposure to many compounds can have adverse consequences, the fact that exposure to addictive drugs is self-selected is a unique feature of their toxicity. The toxic effects of cocaine can be divided into three general categories—motivational toxicity which refers to the effects of the drug on the organization of the animal’s behavior, pharmacological toxicity which concerns the direct pharmacological effects of the compound, and physiological toxicity which includes both neuro-adaptive responses and possible neurotoxic effects following repeated cocaine exposure. Motivational Toxicity A new consideration in the toxicological profile of a substance is its potential motivational toxicity. Addictive drugs can severely disrupt the motivational hierarchy of the organism and produce a type of toxicity conceptually similar to classically defined pharmacological and physiological toxicity. The adverse consequences of drug exposure are not viewed in relation to their direct effects on biological processes, but rather in relation to their adverse behavioral consequences that can severely threaten the survival of the organism. Motivational toxicity can have adverse biological consequences that result from the drug’s influence of on the organism’s behavior. In laboratory animals this may be expressed as a disruption in the regulation of food and water intake, and in humans this effect may be manifest as an increase in risk taking and a disruption of social and legal constraints on behavior. (See Wise and Bozarth, 1985, for a discussion of motivational toxicity; see also Johanson et al., 1987.) Pharmacological Toxicity Intravenous cocaine self-administration in laboratory animals usually involves testing the subjects for several hours per day. Under these conditions of limited drug access, regular patterns of cocaine intake quickly emerge (see Yokel, 1987), and the animals maintain good general health with very few fatalities. Tests involving continuous, 24-hour access to intravenous cocaine have a much different outcome. Animals allowed unlimited access to intravenous cocaine show erratic patterns of self-administration and severe toxic effects (Bozarth and Wise, 1985; Deneau et al., 1969; Johanson, Balster, and Bonese, 1976; Pickens and Thompson, 1971). In a study with laboratory rats, 60% of the animals intravenously self-administering cocaine died by the end of two weeks of testing, while only 9% of the subjects given unlimited access to intravenous heroin had died (see Figure 2). Human cocaine abuse is not associated with a particularly high fatality level, but this is probably because human users generally do not have access to unlimited supplies of cocaine and because a relatively ineffective method of drug administration (i.e., intranasal) is usually used. Increases in the availability of cocaine and the widespread use of more effective routes of administration (i.e., intravenous, smoke inhalation) may dramatically increase the incidence of human fatalities reported with cocaine use. Physiological Toxicity In addition to direct pharmacologically toxic effects, repeated cocaine administration may produce physiologically toxic effects related to its actions on neurotransmitter systems. Although these effects also involve pharmacological mechanisms, adaptive changes in the neural systems affected by the drug are critical for these actions. Specifically, physiological changes may occur that produce impaired neural functioning following repeated cocaine administration. The physiological toxicity considered in this section can be divided into two types—short term (viz., withdrawal) and long-term changes in neurotransmitter systems. Putative Cocaine Withdrawal Behavior Psychomotor stimulants do not produce the type of withdrawal syndrome characteristic of opiate dependence in humans and animals. High doses of opiates generally suppress behavior and autonomic functioning; the withdrawal from these drugs is associated with exaggerated behavioral and autonomic responses, such as escape behavior, teeth chattering, and diarrhea in animals (Bläsig, Höllt, Herz, and Paschelke, 1976; Way, Loh, and Shen, 1969). Stimulants, on the other hand, produce marked behavioral and autonomic stimulation, and it is likely that their withdrawal is associated with a decrease in these systems. That is, while opiate withdrawal is quantified by the appearance of certain behaviors that have a low frequency of occurrence in normal animals, stimulant withdrawal may consist of a decrease in behaviors that have a high frequency of appearance in normal animals. A preliminary test of this hypothesis has been completed (M. Bozarth and M. Hamilton, unpublished observations). Adult male rats were intraperitoneally injected with cocaine (40 mg/kg/day) for 7 days. After 48 hours of drug abstinence, the rats were tested for their exploratory activity using an open field. An automated apparatus was used that measured the number of squares (20 x 20 cm) crossed during a 30 minute test. The open field consisted of 25 grids arranged in a 5 x 5 matrix in a 1.2 m2 enclosure. The primary variable studied was the total number of entries into each grid during the test, and open field activity was measured every other day for the 12 days. Testing occurred during the light phase of a 12-hour light/dark cycle of illumination. There was a significant decrease in open field activity in animals treated with cocaine (see Figure 3). The depression in motor activity abated over the 12 days of testing, indicating that exploratory behavior recovered following drug abstinence. Control rats injected with physiological saline showed little variation in their performance in the open field with repeated testing. These data suggest that an easily quantifiable withdrawal effect does accompany termination of repeated cocaine administration. The behavioral effect consisted of a decrease in locomotor activity—a diminished frequency of occurrence for a behavior normally present. This finding fits well with classical explanations of the neural mechanisms of withdrawal responses (e.g., Himmelsbach, 1943; Jaffe and Sharpless, 1968). Withdrawal reactions are usually the opposite of the direct pharmacological actions of the dependence-producing substance. For example, high doses of opiates produce decreases in motor activity and hypothermia. Opiate withdrawal is usually accompanied by an increase in motor activity and hyperthermia—effects that are directly opposite those produced by the pharmacological action of opiates. Cocaine produces a marked stimulation of locomotor activity and exploratory behavior. The withdrawal effect of this drug, based on the experience with opiates, would be predicted to be a decrease in locomotor activity. The present study provides tentative confirmation of this hypothesis. Furthermore, decreased exploratory behavior may provide an animal model of the "crash" and Phase 1 cocaine abstinence syndrome (see Gawin and Kleber, 1986) that follows high-intensity cocaine usage in humans. Chronic Stimulant Effects on Brain Function The effects of long-term stimulant administration has been an area of considerable research. A number of laboratories have documented the effects of chronic amphetamine treatment on electrophysiological (Ellinwood and Lee, 1983; White and Wang, 1984) and neurochemical (Ellison, 1983; Ellison, Eison, Huberman, and Daniel, 1978; Finnegan, Ricaurte, Seiden, and Schuster, 1982; Lynch, Kenny, and Leonard, 1977; Nwanze and Jonsson, 1981; Preston, Wagner, Schuster, and Seiden, 1985; Schmidt, Ritter, Sonsalla, Hanson, and Gibb, 1985; Seiden, Fischman, and Schuster, 1975/76) measures of brain dopamine function. With repeated amphetamine administration, there is a decrease in the sensitivity of dopamine neurons to the electrophysiological effects of dopamine agonists such as apomorphine (Ellinwood and Lee, 1983). Levels of catecholamines and their metabolites show a marked reduction in terminal field areas, and some investigators have even reported a loss of dopamine cells (Ricaurte, Seiden, and Schuster, 1984). Changes in dopamine-receptor binding have also been reported (Schmidt et al., 1985). These effects are usually present without signs of gross pathological disturbances in the brain (Schuster and Fischman, 1975; Shybut, Richter, and Schuster, 1976) but are associated with a pronounced augmentation of the behavioral effects of stimulants (Ellinwood, Stripling, and Kilbey, 1977; Kilbey and Ellinwood, 1977; Segal and Mandell, 1974; Vogel, Miller, Waxman, and Gottheil, 1985). Progressive increases in the sensitivity of the animals to stimulant drugs are usually noted, although the relative importance of physiological and conditioning factors has not been established (see Demellweek and Goudie, 1983). The depletions of catecholamines are long lasting, persisting for at least several months following the termination of amphetamine injections (e.g., Ellinson, 1983). There is some evidence of a recovery to normal levels of neurotransmitters, but other evidence suggests enduring alterations in brain function and behavior. Because the ventral tegmental dopamine system is important in stimulant reward and because it appears to have a role in the regulation of other motivated behaviors, the effect of chronic stimulant self-administration on dopamine systems is very important. If the function of this brain system were disrupted by chronic stimulant self-administration, then alterations in motivation would be likely to occur. Also, because dopaminergic mechanisms appear to be involved in certain types of psychopathology (e.g., Ellinwood and Kilbey, 1977), disturbances in cognition and affect may also be present. Changes in Monoamines Following Cocaine Self-Administration The studies reporting alterations in brain dopamine systems are all based on experimenter-delivery of high drug dosages. In addition, most studies use amphetamine (e.g., Ellison, 1983; Finnegan et al., 1982; Nwanze and Jonsson, 1981; Schmidt et al., 1985), although at least one study has reported a failure to detect altered dopamine levels following repeated cocaine administration (Seiden and Kleven, 1988). A preliminary study has been completed that evaluates the effects of self-administered cocaine on brain dopamine systems (M. Bozarth, unpublished observations; see also Bozarth, 1986b). The critical variable that distinguishes this study from earlier work is that the drug was self-administered. In this way, drug levels are not arbitrarily selected by the experimenter, but rather reward-relevant dosages are used. The animals determine how much drug is taken as well as the pattern of drug intake. This experimental approach assesses if rewarding dose-levels of cocaine cause alterations in brain dopamine systems. It is, therefore, directly relevant to cocaine abuse because the important variable of drug administration is controlled by the subjects and it does not focus on arbitrarily selected, high drug dosages. Furthermore, because cocaine abuse is currently at a high level, this drug was examined rather than amphetamine. Even though similar neurochemical effects are likely to occur, the failure to see documented evidence of long-term catecholamine depletions following cocaine treatment suggests that there have been problems in establishing this effect. Male, Long-Evans rats received intravenous catheters and were tested for cocaine self-administration. Rats were tested using a continuous reinforcement schedule where each lever press delivered a 1 mg/kg/injection of cocaine over 28 seconds. Testing lasted for 2 hours per day, and animals were tested 5 days per week for a total of 3 weeks. Levels of cocaine self-administration were similar to those seen with other studies in this laboratory using the same testing protocol (i.e., mean intake [approx.] 6 mg/kg/hr). Seventy-two hours after the last cocaine self-administration session, the rats were decapitated and the brains rapidly removed and frozen for later analysis. Serial, 300 micron coronal brain sections were taken at areas corresponding to the major ventral tegmental and nigrostriatal dopamine terminal fields and cell body regions. These sections were placed on slides, and small tissue samples corresponding to areas in these dopamine systems were removed using a micropunch. The tissue samples were then assayed for monoamines and their major metabolites using high performance liquid chromatography with electrochemical detection. Comparisons were made with untreated control rats, and all samples were equated based on protein content of the micropunched regions. There was a marked reduction in dopamine content in the nucleus accumbens and ventral tegmental area (see Figure 4). Slight reductions were also seen the dorsal striatum and substantia nigra, but these changes were not very large. The fact that major reductions in dopamine levels were not seen in other dopamine-containing brain regions suggests that these depletions were relatively specific to the ventral tegmental dopamine system. An analysis of monoamine levels in the nucleus accumbens showed that in addition to dopamine, serotonin levels were also decreased (see Figure 5). There was no significant change in norepinephrine, indicating that the effects of intravenous cocaine self-administration were limited to dopaminergic and serotonergic levels in the systems studied. The result of this preliminary study suggests that intravenous cocaine self-administration for even a relatively short period of time produces marked decreases in nucleus accumbens dopamine and serotonin. The decrease in dopamine levels is similar to that reported for experimenter-delivered amphetamine and is likely to influence the rewarding effects of cocaine. The decrease in serotonin content is interesting because this neurotransmitter may be related to depression and other affective changes following termination of cocaine self-administration. Implications for Clinical Treatment Animal research has contributed substantially to an understanding of the rewarding, and hence addictive, properties of drugs. The demonstration that physical dependence is not a necessary attribute of drug reward causes a reassessment of the widely held definition of addiction. Drug addiction is seen as an operant behavior, much like any other behavior, that has developed to a compulsive level. The fact that animals self-administer drugs without any necessary predisposing conditions illustrates that some drugs can serve as universal reinforcers and predisposing psychological factors are not necessary for an addiction to occur. There are two important implications from these observations applicable to clinical practice. First, physical dependence is probably a concomitant of addiction to some drugs and not a primary determinant in the etiology of drug addiction. Second, personality traits are likely to contribute to the initial acquisition of drug-taking behavior (viz., tendency to experiment with drugs and social acceptability of drug use), but they are not likely to be controlling factors in the development of drug addiction. Rather, they significantly influence initial drug exposure (albeit a necessary part of drug addiction), but potent pharmacological effects are likely to account for the full development of the drug addiction. Dopaminergic mechanisms in the nucleus accumbens are critically involved in the rewarding effects of cocaine. Neuroleptics should block the rewarding effects of cocaine and thus could provide a pharmacological treatment for cocaine addiction. There are several reasons, however, that neuroleptics are not a viable treatment for cocaine addiction. First, the pharmacological side-effects of these drugs limit the therapeutic dosages that can be used. Because neuroleptics are competitive antagonists at the dopamine receptor, increasing the amount of cocaine injected can overcome the neuroleptic blockade. The cocaine user need only increase his dosage to produce the desired pharmacological effect. This necessitates the use of very high neuroleptic doses in ambulatory patients, and severe side-effects are very likely to occur with chronic neuroleptic treatment. Second, there are likely to be motivational side-effects of neuroleptic treatment. Most notable is anhedonia which involves a blunting of the rewarding impact of motivational stimuli by neuroleptic administration (see Wise, 1982). The ability of other, natural rewards to control the individual’s behavior is likely to be disturbed and psychological depression may accompany chronic, high-dose neuroleptic treatment. Third, neuroleptic blockade of the pharmacological reward from cocaine administration does not necessary block the motivation for cocaine. Although behavioral extinction may occur with repeated cocaine administration during neuroleptic treatment, compliance with neuroleptic treatment may be very difficult to obtain in most patients. The use of long-acting depot neuroleptics may seem to circumvent this problem, but treatment dropout rates are likely to remain high unless drug craving can be abated. If long-term changes in dopamine systems occur following repeated cocaine use in humans, a new dimension is added to the problem of cocaine addiction. Preliminary animal data reported in this paper suggest an impaired dopaminergic and serotonergic functioning following intravenous cocaine self-administration. Enduring deficits in a reward substrate are likely to create another motivational effect of cocaine. Neuroleptic treatment in this situation would be expected to further increase the motivation for cocaine and is thus contraindicated. Pharmacological treatments aimed at restoring normal dopaminergic function may contribute significantly to remission in cocaine addiction (see Dackis and Gold, 1985). Pharmacological intervention is likely to be very important if efficacious treatments can be developed. Preliminary results using the dopaminergic agonist bromocriptine (e.g., Giannini and Billett, 1987; Roehrich, Dackis, and Gold, 1987) and the antidepressant desipramine (e.g., Gawin, Kleber, Byck, Rounsaville, Kosten, Jatlow, and Morgan, 1989) are very encouraging. The potential usefulness of these compounds might have been predicted from earlier animal studies. The rewarding effect of direct electrical brain stimulation appears to be dependent on the activation of a brain dopamine system (Crow, 1972; Fibiger, 1978; Wise, 1978; see also Bozarth, 1987c). Dopamine-depleting lesions disrupt responding for brain stimulation reward, and bromocriptine can restore responding for the rewarding stimulation (Carey, 1983). Also, chronic treatment with desipramine enhances the rewarding effects of electrical stimulation, while acute treatment—associated with noradrenergic effects—does not (Fibiger and Phillips, 1981). Both findings are consistent with the notion that these compounds may help to restore dopamine function in the brain reward system involved in cocaine addiction. The prognosis for recovery from cocaine addiction is probably very poor. Abstinence from the drug is obviously important, and avoidance of stimuli associated with cocaine use is likely to help diminish drug craving. Alternatively, drug craving may extinguish if the patient is exposed to drug-related stimuli without receiving cocaine (e.g., Childress, Ehrman, McLellan, and O’Brien, 1988). Treatment approaches designed to diminish drug craving through a combination of behavioral extinction and pharmacological intervention (e.g., O’Brien, Childress, Arndt, McLellan, Woody, and Maany, 1988) are likely to improve success rates. The prognosis should improve markedly with the duration of abstinence. However, considerable basic and preclinical research is needed before an effective treatment for cocaine addiction will be available.