R. Spanagel and F.

Weiss – Dopamine and reward REVIEW

The dopamine hypothesis of reward:

past and current status
Rainer Spanagel and Friedbert Weiss

Mesolimbic dopaminergic neurons are thought to serve as a final common neural pathway for
mediating reinforcement processes.However,several recent findings have challenged the view that
mesolimbic dopamine has a crucial role in the maintenance of reinforcement processes, or the
subjective rewarding actions of natural rewards and drugs of abuse. Instead, there is growing
evidence that dopamine is involved in the formation of associations between salient contextual
stimuli and internal rewarding or aversive events.This evidence suggests that dopaminergic-neuron
activation aids the organism in learning to recognize stimuli associated with such events. Thus,
mesolimbic dopaminergic neurons have an important function in the acquisition of behavior
reinforced by natural reward and drug stimuli. Furthermore, long-lasting neuroadaptive changes
in mesolimbic dopamine-mediated transmission that develop during chronic drug use might
contribute to compulsive drug-seeking behavior and relapse.
Trends Neurosci. (1999) 22, 521–527

E XTERNAL AND INTERNAL STIMULI have signaling

and reinforcing functions. In its reinforcing capacity,
a stimulus increases or decreases the frequency of pre-
transmission consistently, despite the fact that these
drugs have rewarding properties and are heavily
abused8–10.
ceding responses, and accordingly is called a reinforcer The responses to conditioned reinforcers – that is,
or, respectively, a punisher. In 1954 experiments by those stimuli that gain their reinforcing efficacy by
Olds and Milner1 revealed that the brain has specialized means of their association with unconditioned reward-
‘centers’ for reward functions. In these studies, electrical ing natural or drug stimuli – have also been linked to
stimulation of specific brain sites was found to be highly midbrain dopamine-mediated transmission. However,
rewarding in the sense that rats responded operantly to conditional changes in dopamine release in the NAC
electrical stimulation of these brain sites, often to the ex- are found only after stimuli that are predictive of strong
clusion of any other activity. A neurotransmitter system aversive events such as footshock11 or, in the case of ap-
that is particularly sensitive to electrical self-stimulation petitive stimuli, under food-deprivation conditions but
is the midbrain dopaminergic projection that originates not under ad libitum conditions12,13. In addition, pref-
in the ventral tegmental area (VTA) and projects to erential activation of midbrain dopaminergic neurons
structures closely associated with the limbic system, most by appetitive rather than aversive stimuli has been re-
prominently the nucleus accumbens (NAC) shell region ported in monkeys14, suggesting that conditioned stim-
and the prefrontal cortex (Fig. 1). Because of its ubiqui- uli can enhance mesolimbic dopamine release, but that
tous involvement in the regulation of reward-related this effect depends on the type of the unconditioned
behavior, this system has been characterized as a neuro- stimuli used and on the deprivation state.
chemical substrate of reward2. Synaptic dopamine-
Natural rewards and midbrain dopamine
mediated transmission in the NAC shell region is in-
creased preferentially not only by natural rewards such Numerous studies have shown that dopamine release
as food, water and sex, but also by a variety of drugs in the NAC is increased by ingestion of food and water.
abused by humans3,4. Furthermore, it has been hypothe- The observation that microinfusion of the m-opioid-
sized that long-term adaptive changes within midbrain receptor antagonist, naloxonazine, into the VTA impairs
dopaminergic neurons take place following repeated food-induced stimulation of dopamine release in the
drug administration and lead to a disruption or desen- NAC shell region15, indicates that food acts via a specific
sitization of neural mechanisms that mediate reward5,6, neuronal mechanism in this brain region that leads to
while rendering these neurons more sensitive to other an activation of dopaminergic neurons (Fig. 2). However, Rainer Spanagel is
behavioral actions of drug stimuli7. consistent stimulation of dopamine release by food or at the Max Planck
liquids can be demonstrated only under deprivation Institute of
Do all natural rewards and drugs of abuse interact
conditions and not in nondeprived rats12,13, suggesting Psychiatry, 80804
with midbrain dopamine?
a role for incentive rather than consummatory factors Munich, Germany,
Natural rewards, including food, liquids and sex, as in dopamine-mediated activation by natural rewards. and
well as electrical brain stimulation and many drugs of Another major determinant of the responsiveness of Friedbert Weiss
abuse, increase extracellular dopamine levels in the NAC midbrain dopamine to natural rewards appears to be is at the Scripps
following acute administration2–4. However, many novelty. Thus, in rats fed ad libitum with standard food, Research Institute,
agents, such as inhalants, barbiturates or benzodi- feeding of a novel, palatable food elicits an immediate La Jolla,
azepines, do not activate midbrain dopamine-mediated increase in extracellular dopamine release in the NAC, CA 92037, USA.

0166-2236/99/$ – see front matter © 1999 Elsevier Science Ltd. All rights reserved. PII: S0166-2236(99)01447-2 TINS Vol. 22, No. 11, 1999 521
REVIEW R. Spanagel and F. Weiss – Dopamine and reward

between conditioned and unconditioned stimuli, and

detecting the occurrence versus omission of an expected
Rewards Rewards reward16 (see Box 1).
D1 and D2
VTA
Psychostimulants and midbrain dopamine
receptors
Both amphetamine and cocaine elevate extracellular
A10 dopamine concentrations in the terminal region of mid-
brain dopaminergic neurons, especially in the NAC
− Interneuron (Refs 3,4,8). Extracellular dopamine levels are also re-
liably increased during self-administration of these
−
NAC drugs17,18. However, psychostimulant-induced increases
Rewards Disinhibition in synaptic dopamine in self-administering animals
Long-loop GABAA receptors might involve factors other than only the direct phar-
GABAergic macological actions of these drugs. Rats that received
feedback trends in Neurosciences non-response-contingent injections of cocaine admin-
istered by an animal self-administering cocaine, by
Fig. 1. The mesolimbic dopamine system. A10 dopaminergic neurons
means of a ‘yoking’ procedure, showed substantially
originate in the ventral tegmental area (VTA) and project mainly to the
nucleus accumbens shell region (NAC). GABAergic interneurons within
smaller increases in extracellular dopamine concen-
the VTA and a long-loop GABAergic feedback projection provides tonic trations than self-administering rats19. Thus, as suggested
inhibition (2) of A10 neurons. Different dopamine receptors (D1 and D2) by the differences in the effects of food in deprived
are involved in mediating the action of rewards. versus nondeprived animals on dopamine release, the
‘incentive salience’ of the stimulus (incentive salience
but this effect is blunted by a previous meal of the refers to the magnitude of a reinforcing stimulus)
palatable food13. These results indicate that novelty is appears to have an important role in the response of
important for activation of mesolimbic dopamine by midbrain dopaminergic neurons to rewarding stimuli.
food stimuli in a nondeprived condition, and that the Experiments that examined the dopaminergic response
dopamine signal adapts to the repeated presentation of to cocaine in self-administering rats with high time res-
the same natural reward. Electrophysiological studies olution suggest that the cocaine response is regulated by
in monkeys further illustrate the role of midbrain do- changes in extracellular dopamine levels in the NAC
paminergic neurons in the processing of information re- (Refs 20–22). Self-injections of cocaine produce a tran-
lated to natural rewards, and suggest that these neurons sient rise in dopamine levels; when synaptic dopamine
have an important role in the learning of stimulus– concentrations decrease to preinjection levels, the
reward associations by coding the temporal contiguity next lever press is initiated. Thus, phasic changes in

A B

Food, liquids and sex

(via sensory inputs)
VTA VTA Opioids

− −

− −
NAC NAC
Opioids
µ-Opioid receptors

C D
Cocaine and
amphetamine

VTA Alcohol VTA

DAT

− −

NAC − −
NAC
Alcohol
trends in Neurosciences

Fig. 2. The mechanisms of action of natural and drug rewards on mesolimbic dopaminergic neurons. Natural rewards (food, liquids and sex)
act in an indirect manner to activate A10 neurons (A). Opioids and alcohol also increase dopamine-mediated neuronal activity in an indirect manner
(B) and (C), whereas psychostimulants (cocaine and amphetamine) activate dopamine release by stimulating the release or inhibiting the dopamine
transporter (DAT) (D). Abbreviations: NAC, nucleus accumbens shell region; VTA, ventral tegmental area.

522 TINS Vol. 22, No. 11, 1999

 R. Spanagel and F. Weiss – Dopamine and reward REVIEW

Box 1. Past and current opinions: what is the role of dopamine

within the mesolimbic system?
The hypothesis that mesolimbic dopaminergic neurons associated predominantly with mesolimbic dopamine-
are the final common pathway for positive reinforcement mediated transmission, which often develops with inter-
by natural rewards or drugs of abuse has guided the field mittent exposure to drugs of abuse. However, chronic drug
for more than two decades. Except in the case of psycho- use can also result in dysfunction or desensitization of neur-
stimulant reinforcement, however, this hypothesis remains onal mechanisms that mediate reward, including dopamine
to be confirmed unequivocally, and alternative views on the neurotransmissionf,g. As a result of this compromised con-
role of mesolimbic dopamine-mediated transmission have dition of the reward system, affective states emerge that are
emerged. Recent work by Schultz and co-workers indicates opposite to the initial mood-elevating effects of drugs, such
that, whereas midbrain dopaminergic neurons respond to as dysphoria, depression and anxiety. This view of depend-
natural rewards such as food and liquid, the activation of ence identifies a single motivational consequence of with-
these neurons depends on the predictability of reward drawal, negative affect, rather than physical withdrawal
presentationa. An unexpected reward elicits a strong posi- symptoms in compulsive and escalating drug-seeking be-
tive dopamine signal, which declines with repeated pres- havior, and suggests a common basis for withdrawal from
entation and learning, such that the presentation of a many, if not all, classes of drugs of abuseg.
predicted reward eventually no longer elicits dopamine- Another hypothesis, proposed by Di Chiarah, attributes an
mediated neuronal responses. By contrast, omission of a important role to mesolimbic dopamine-mediated trans-
predicted reward leads to suppression of the dopamine sig- mission in reward-related learning processes, and suggests
nal. These findings suggest that the response of midbrain that drug addiction is a dopamine-dependent associative-
dopaminergic neurons represents a learning signal that learning disorder. This view distinguishes between associa-
codes for prediction errors of rewardb. An alternative inter- tive learning and dopaminergic-neuron activation induced
pretation of these data has been offered, which proposes by natural rewards, and ‘abnormal’ associative learning and
that the dopamine signal facilitates the reallocation of lim- dopamine activity stimulated by drugs of abuse. According
ited behavioral and cognitive processing capacity towards to this hypothesis, activation of dopamine in the NAC shell
any unexpected event of behavioral significance, including region by natural rewards underlies habituation. In con-
reward. This behavioral-switching hypothesisc stipulates trast, drugs of abuse have nonhabituating effects, resulting
that the dopamine signal has a more-general role in asso- in nonadaptive or even sensitized dopamine release after
ciative learning. Indeed, it has been demonstrated that repeated drug useh. These neurochemical consequences of
associative learning is linked to enhanced dopamine release drugs of abuse are thought to strengthen drug stimulus–
in the nucleus accumbens (NAC)d. Thus, a selective increase reward associations, which constitute the basis of addictive
in dopamine release in the NAC, but not in the dorsal stri- behavior.
atum, is seen when an association is formed between two
stimuli, of which neither is a biological reinforcer nor affects References
a Schultz, W. (1997) Curr. Opin. Neurobiol. 7, 191–197
dopamine levels prior to formation of the associationd. b Hollerman, J.R. and Schultz, W. (1998) Nat. Neurosci. 1,
These new findings suggest a role for mesolimbic dopamine 304–309
in the modulation of associative learning in general, not c Redgrave, P., Prescott, T.J. and Gurney, K. (1999) Trends
only that involving reinforcement (see Fig. I). Neurosci. 22, 146–151
With respect to chronic drug use, neuroadaptive pro- d Young, A.N.M. et al. (1998) Neuroscience 83, 1175–1183
e Robinson, T.E. and Berridge, K.C. (1993) Brain Res. Rev. 18,
cesses within the dopaminergic system have been identi-
247–291
fied. Sensitization has been implicated in the development f Wise, R.A. (1996) Curr. Opin. Neurobiol. 6, 243–251
of compulsive drug usee, and involves a dramatic augmen- g Koob, G.F. and Le Moal, M. (1997) Science 278, 52–58
tation of behavioral and neurochemical responses that are h Di Chiara, G. (1998) J. Psychopharmacol. 12, 54–67

Preconditioning Aversive Conditioned responses

conditioning
Associative
learning
− +

(Shock)
+ +

DA− DA− DA↑ DA↑ DA↑ DA↑

trends in Neurosciences

Fig. I. The role of dopamine in associative learning processes. Brain microdialysis was used to study changes in dopamine (DA) in the
nucleus accumbens during associative learning between two neutral stimuli (flashing light and tone)d. Separately, the flashing light or tone did
not affect dopamine release (DA2). However, when presented on a paired schedule during preconditioning, dopamine release was increased
(DA↑). The tone was subsequently paired with mild footshock (aversive conditioning), and the formation of a conditioned association
between the flashing light and the tone was assessed by measuring the ability of the flashing light to elicit the same conditioned response
as the tone when presented during the test.

TINS Vol. 22, No. 11, 1999 523

REVIEW R. Spanagel and F. Weiss – Dopamine and reward

Box 2.The human midbrain dopamine system and the action of drugs of abuse

Cocaine euphoria and reward are linked to dopamine ture suggests that it is involved in both reinforcement and incentive
transporter (DAT) occupancy functions. The stronger association of signal changes in the NAC with
The reinforcing and subjective, pleasurable effects of cocaine are ‘craving’ rather than ‘rush’ ratings (see Fig. I) challenges the view that
thought to depend on blockade of the DAT by cocaine, which results dopamine transmission in the region of the NAC has a central and
in increased synaptic availability of dopamine. A direct relationship selective role in the mediation of reward and subjective euphoria.
between DAT blockade and self-report measures of ‘high’ was recently Cocaine addicts show reduced euphoria and striatal
demonstrated in humansa. The degree and timecourse of DAT block dopamine response to dopamine reuptake inhibitor
by cocaine (as measured by PET) was correlated significantly with the Volkow et al. compared the responses of cocaine addicts and
temporal profile of cocaine euphoria and the binding kinetics of normal controls to intravenous methylphenidatec, a drug that, like
cocaine in the striatum of cocaine abusers. Although these measures cocaine, increases synaptic dopamine levels. The effects of methyl-
were made in the dorsal striatum rather than the nucleus accum- phenidate on both striatal dopamine release (as determined by PET
bens (NAC; the structure that is commonly associated with cocaine measures of [11C]raclopride displacement by endogenous dopamine)
reinforcement in animals), DAT occupancy by cocaine is sufficiently and subjective feelings of ‘high’ were significantly reduced in cocaine-
similar in these brain regions that these results provide evidence for dependent subjects. These findings are compatible with a decrease or
DAT blockade as a crucial mechanism in mediating the reinforcing desensitizaton in striatal dopamine function and a decrease in the
and subjective effects of cocaine in humans. reinforcing efficacy of psychostimulants, but contrast with the view
Both cocaine euphoria and cocaine craving are associated that sensitization (that is, enhanced dopamine transmission or the
with functional MRI (fMRI) activation in the nucleus reinforcing effects, or both) is a mechanism in human cocaine
accumbens addiction.
Breiter et al.b used fMRI in conjunction with behavioral ratings to High doses of a dopamine-receptor antagonist do not block
investigate brain circuitries mediating cocaine-induced euphoria and amphetamine-induced euphoria in humans
craving in cocaine-dependent subjects. Following an infusion of The dopamine hypothesis of reward rests, in part, on evidence that
cocaine, several brain regions, including the ventral tegmentum and dopamine antagonists reliably attenuate the reinforcing actions of
basal forebrain, showed an immediate but transient activation that psychostimulants in animals. However, in studies designed to exam-
was correlated with reinforcement-related ratings of ‘rush’. Other ine whether the role of dopamine in psychostimulant reward in
regions, including the NAC and amygdala, showed sustained acti- animals extends to psychostimulant-induced euphoria in humans,
vation that was closely associated with incentive-related measures the dopamine antagonist, pimozide, failed to antagonize consistently
of craving. Although some rush-associated early activation was ob- the euphorigenic effects of amphetamined. By failing to demonstrate a
served in the NAC, the temporal pattern of activation in this struc- role for dopamine in the mood-elevating effects of amphetamine, these
data call into question the assumption that dopamine-dependent
reinforcing effects of psychostimulants in animal models are directly
related to drug-induced euphoria in humans.
Neuroadaptations in the dopamine system in drug addicts
In humans, cocaine withdrawal is associated with persistent hyper-
prolactinemiae, a condition that might reflect reduced dopaminergic-
neuron function because prolactin secretion is inhibited tonically by
hypothalamic dopamine. More-direct evidence for an enduring do-
paminergic-neuron hypofunction comes from brain-imaging studies
showing that striatal dopamine synthesis decreases with increasing
duration of abstinencef. A hypofunctioning of the dopaminergic sys-
tem has also been reported following long-term alcohol abuse.
Alcohol-abuse patients show impairments in the function of central
dopamine receptors that depend on the amount of long-term alco-
hol consumption, and retardation in the recovery of dopamine
function is associated with early relapse in alcoholicsg.

References
Fig. I. Functional MRI (fMRI) during cocaine craving. Brain regions demonstrate a Volkow, N.D. et al. (1997) Nature 386, 827–830
significant activation in response to viewing a film depicting explicit cocaine use b Breiter, H.C. et al. (1997) Neuron 19, 591–611
in a group (n=14) of experienced cocaine users. A self-report questionnaire revealed c Volkow, N.D. et al. (1997) Nature 386, 830–833
d Brauer, L.H. and DeWit, H. (1997) Pharmacol. Biochem. Behav. 56,
high scores of craving in most of the cocaine users compared with controls. Colors
265–272
represent percentage change of fMRI signal averaged over 7 mm of activated cor- e Teoh, S.K. et al. (1990) Biol. Psychiatry 28, 824–828
tex. Depicted regions include the bilateral superior frontal gyrus (A), cingulate gyrus f Wu, J.C. et al. (1997) Neuropsychopharmacology 17, 402–409
and gyrus rectus (B), and the bilateral amygdala (C). Data courtesy of E. Stein. g Heinz, A. et al. (1996) Arch. Gen. Psychiatry 53, 1123–1128

dopamine levels might be a mechanism that trans- knockout and wild-type mice to self-administer cocaine
duces declining brain cocaine levels into drug-seeking intravenously. Contrary to expectations, DAT-knockout
behavior22. These observations extend earlier findings mice acquired self-administration of cocaine. Thus,
showing that selective destruction of midbrain dop- the reinforcing actions of cocaine did not depend on
amine neurons by 6-hydroxydopamine (6-OHDA) cocaine-induced increases in synaptic dopamine in these
abolish cocaine self-administration23, and support the mutants, because cocaine injections markedly increased
hypothesis that the mesolimbic dopaminergic system is extracellular dopamine levels in wild-type mice but did
crucial for psychostimulant reward. not alter the already high synaptic dopamine levels in
As cocaine-induced increases in extracellular dop- the DAT knockouts24. The observation of specific bind-
amine concentrations are due to the block of presynaptic ing of a cocaine analog and enhanced Fos expression
dopamine transporters (DAT), disruption of the DAT in response to cocaine in serotonergic brain regions
should attenuate the reinforcing effects of cocaine. In of the DAT-deficient mutants24 led to speculation that
order to test this hypothesis, Rocha et al.24 trained DAT the 5-HT transporter might provide a mechanism for

524 TINS Vol. 22, No. 11, 1999

 R. Spanagel and F. Weiss – Dopamine and reward REVIEW

the reinforcing actions of cocaine in these animals. TABLE 1. Dopamine release in the nucleus accumbens
However, while occupancy of 5-HT transporters might in response to natural and drug rewards
be sufficient to initiate self-administration of cocaine
in DAT knockouts, this hypothesis was not supported Reward Acute Repeated Withdrawal
by a recent study in which conditioned place preference (sensitization)
was established in DAT as well as in 5-HT-transporter
Natural reward ↑ 2 2
knockout mice25. Thus, neither the DAT nor the 5-HT Opioids ↑ ↑↑a ↓
transporters alone, seems essential for conditioned Psychostimulants ↑ ↑↑ ↓
reinforcement associated with cocaine. Although these Alcohol ↑ ↑b ↓
new findings challenge the dopamine hypothesis of
cocaine reward, it is possible that developmental adap- Abbreviations: ↑↑, augmented increase; ↑, increase; ↓, decrease; 2, no
tations in these mutants might supervene and allow change.
a
another transporter to substitute for function of the Mixed results27,28.
b
deficient transporter. One candidate might be the nor- No augmented increase and no tolerance29,30.
adrenaline transporter. A recent study suggests that
cocaine, as well as amphetamine, increases extracellular heroin self-administration is mediated by dopamine-
dopamine levels largely through inhibition of the nor- independent mechanisms, presumably opiate receptors
adrenaline transporter, although these results were ob- localized postsynaptically in the NAC (Ref. 8).
tained in the prefrontal cortex26. In addition, compen- In contrast to self-administration, opiate reward, as
satory adaptations in these mutants could result in measured by the conditioned place-preference method,
upregulation of as yet unknown pharmacological tar- seems to depend on midbrain dopamine-related mecha-
gets of cocaine that might not have a significant role in nisms. Microinjections of m-opioid-receptor agonists into
wild-type animals. These issues might soon be resolved the VTA, but not NAC, induced conditioned place pref-
by the use of inducible monoamine-transporter knock- erence39,40. This effect could be blocked by injection of
outs. The involvement of monoamine transporters and 6-OHDA or the dopamine antagonist SCH23390 into
dopamine in cocaine reinforcement has been also stud- the NAC (Ref. 41). However, D1 receptors are not the
ied using various noninvasive neuroimaging approaches only receptors that are crucial for morphine-induced
in humans (see Box 2). However, although these ex- conditioned place preference. In mice that lack D2 re-
periments confirm a role for the DAT transporter and ceptors, suppression of morphine-induced place pref-
activation of dopamine-rich brain regions in the sub- erence was observed42. These data suggest that the
jective effects of cocaine, the view that mesolimbic acquisition and expression of the secondary reinforc-
dopamine-mediated transmission has a central and ing effect of opioids, as measured by conditioned place
selective role in the mediation of psychostimulant preference, depend on the functional integrity of
euphoria has not remained unchallenged in human mesolimbic dopaminergic neurons, whereas dopamine-
experimental studies (see Box 2). independent mechanisms have to be postulated for me-
diating primary reinforcing effects of opioids during
Opioids and midbrain dopamine
self-administration.
Opiate drugs and opioid peptides can modulate meso-
Alcohol and midbrain dopamine
limbic dopamine activity (Table 1). In vivo microdialysis
data demonstrate that acute systemic or intracerebro- Several lines of evidence indicate that ethanol acti-
ventricular administration of m- or d-opioid-receptor vates the mesolimbic dopaminergic system. Alcohol
agonists increase dopamine release in the NAC, whereas injected intravenously increased firing of dopamine
k-opioid-receptor agonists decrease dopamine release31. neurons in the VTA (Ref. 43), and acute administration
Opioid-receptor agonists increase extracellular dopamine of alcohol resulted in preferential release of dopamine
levels within the NAC by disinhibiting GABA inter- from the NAC shell region3,44. Similar to opioid-induced
neurons in the VTA (Ref. 32). Activation of m-opioid re- stimulation of dopamine release, alcohol is thought to
ceptors on GABAergic interneurons hyperpolarizes these decrease the activity of GABAergic neurons in the VTA,
interneurons and concomitantly disinhibits dopamine- which leads to a disinhibition of mesolimbic dopamin-
cell firing32. These disinhibitory actions of opiate ag- ergic neurons45; however, alcohol might also have some
onists are restricted to the VTA as direct application of local effects in the NAC (Ref. 45). Indeed, the activation
m-opioid-receptor agonists into the midbrain increases of dopaminergic neurons by ethanol might involve an
mesolimbic dopamine-mediated activity, whereas intra- interaction with endogenous opioids in the VTA, as the
NAC infusions do not alter extracellular dopamine lev- suppression of alcohol intake by nonselective opiate-
els in this structure33. In contrast to these effects of non- receptor antagonists has been linked to interference of
contingent opioid-receptor agonist treatments, heroin these agents with the dopamine-stimulatory actions
self-administration failed to elevate extracellular dop- of ethanol46.
amine levels in the NAC (Ref. 34). This finding is con- Numerous pharmacological studies have investigated
sistent with several pharmacological studies showing the role of midbrain dopaminergic neurons in alcohol
that administration of dopamine antagonists, either reinforcement, but the results have been inconsistent29.
systemically35,36 or directly into the NAC (Ref. 37), does Lesioning of dopaminergic neurons by 6-OHDA does not
not alter the responses maintained by intravenous affect the maintenance of alcohol self-administration47,48,
heroin. Furthermore, selective destruction of presynap- whereas acquisition of alcohol drinking is substantially
tic dopaminergic nerve terminals in the NAC, using the reduced by this manipulation48. These findings illustrate
neurotoxin 6-OHDA, does not attenuate intravenous that different neuronal mechanisms mediate acquisition
opiate self-administration37,38. In contrast to the pre- and maintenance of alcohol drinking and that func-
diction of the classical dopamine hypothesis of tional integrity of midbrain dopaminergic neurons is
reward, these data suggest that the maintenance of not required to maintain alcohol self-administration.

TINS Vol. 22, No. 11, 1999 525

REVIEW R. Spanagel and F. Weiss – Dopamine and reward

However, rats will self-administer ethanol directly into attenuate stress-induced reinstatement of heroin self-
the ventral tegmental cell-body region of mesolimbic administration57, a finding that argues against a role for
dopaminergic neurons49, and both D1- and D2-receptor dopamine in stress-associated heroin-seeking behavior,
antagonists administered either systemically or locally in spite of the fact that stress stimuli typically produce
into the NAC decrease home-cage drinking and operant robust increases in accumbal dopamine release. Another
responses to alcohol50,51. study has demonstrated an opposite modulation of
Measurements of ethanol-induced dopamine release cocaine reinstatement by D1- and D2-receptor agonists59,
in genetically selected alcohol-preferring rat strains whereby these agonists were able to induce reinstate-
have also produced conflicting results29. Many alcohol- ment of cocaine-seeking behavior. Moreover, D1-receptor
preferring strains show deficiencies in forebrain do- agonists prevented reinstatement behavior induced by
paminergic function compared with their nonpreferring cocaine priming, whereas D2-receptor agonists even
counterparts, and it has been proposed that these ab- enhanced this behavior59. As inhibition of cAMP-
normalities could be a factor in genetically determined dependent protein kinase in the NAC also reinstated
ethanol preference. However, in alcohol-naïve high- cocaine-seeking behavior60, a role for dopamine in the
alcohol-drinking (HAD) and low-alcohol-drinking (LAD) D1-linked cAMP system in the NAC has been proposed
strains of rat, dose–response curves for alcohol-induced as a potential mechanism in relapse to cocaine-seeking
dopamine release show no difference in sensitivity to behavior. Examination of dopaminergic-neuron func-
alcohol between the lines52. A similar lack of line differ- tion in the NAC associated with the alcohol-deprivation
ences has been reported in alcohol-preferring and alco- effect, which has been proposed as a model for relapse
hol-avoiding rats53,54. Furthermore, in alcohol-preferring and craving58, revealed that heightened alcohol-seeking
rats with a history of alcohol self-administration, dop- behavior was accompanied by enhanced dopamine
amine release after alcohol injections was completely release61. In agreement with this finding, craving induced
attenuated54. In contrast, operant responses to alcohol by alcohol priming in detoxified alcoholics could be pre-
were associated with a considerably greater relative vented by haloperidol pretreatment62. Conversely, ad-
stimulation of dopamine release within the NAC in ministration of dopamine-receptor antagonists did not
alcohol-preferring P-rats than in control Wistar rats55. influence or even enhance relapse in human alcoholics29.
These data suggest that ethanol preference is not related These findings suggest that dopamine-related mecha-
to the amount of dopamine released by noncontingent nisms might have a role in ethanol-priming effects, but
alcohol treatment; however, concomitant measures of not in the resumption of drinking elicited by other
ethanol-maintained reinforcement and dopamine stimuli.
release have revealed a possible link between genetic While repeated administration of many drugs of abuse
ethanol preference and ethanol-induced dopaminergic- can induce sensitized dopaminergic-neuron responses,
neuron activation55. chronic drug administration can also lead to dopamin-
ergic-neuron dysfunction that becomes unmasked dur-
Long-term changes in mesolimbic dopamine
ing withdrawal. Thus, marked inhibition of mesolimbic
release following chronic drug administration
dopamine release seems to be a common feature of
Repeated administration of psychoactive drugs can drug withdrawal in rats63. It is suggested that long-term
have neuroadaptive consequences that lead to either a exposure to a drug suppresses basal dopamine-medi-
decrease (tolerance or desensitization) or an increase ated activity in order to ‘balance’ chronic stimulation
(sensitization) of their behavioral effects. In the case of by this drug. Such dopaminergic-neuron dysfunction
psychostimulants and opioids, in vivo measurements following chronic drug use is not restricted to the acute
of extracellular dopamine levels have provided direct withdrawal phase as reduced activity of dopamine neur-
evidence that these drugs, when administered under ons is still present long after termination of chronic
an intermittent injection schedule, can lead to a more- alcohol treatment (>3 days), although behavioral mani-
pronounced increase in dopamine levels than the in- festations of the alcohol-withdrawal syndrome recede
crease seen after acute administration of these drugs27,28; within several hours64,65. A similar long-lasting decrease
however, rats that received repeated intermittent injec- in mesolimbic dopaminergic-neuron activity has been
tions of alcohol did not show sensitization of alcohol- observed following cocaine withdrawal66.
induced dopamine release in the NAC (Ref. 30). Whether Experimental studies of the behavioral significance of
sensitization of mesolimbic dopamine neurons is linked dopaminergic-neuron dysfunction are, however, sparse.
to enhanced rewarding efficacy remains unclear. Indeed, Withdrawal from many drugs of abuse attenuates the
current conceptualizations of the significance of sensi- rewarding effects of electrical brain stimulation, a phe-
tization in compulsive drug-seeking behavior hold that, nomenon that has been linked to the negative emo-
rather than enhancing ‘reward’, repeated drug use leads tional changes in withdrawal states5. As brain-stimu-
to a progressive and persistent hypersensitivity of neural lation reward is sensitive to disruption of mesolimbic
systems that mediate ‘incentive salience’, resulting in dopamine transmission, withdrawal-associated reward
excessive craving6,56 (see Box 1). Craving elicited by drug deficits might be linked to impairments in mesolimbic
stimuli that share stimulus properties of the abused drug dopamine-mediated transmission5.
(that is, ‘priming effects’) and drug-related conditioned
Concluding remarks
cues is one factor, among others (for example stress),
that can induce relapse57,58. There is now little doubt that midbrain dopamine
There is evidence to suggest that dopamine-related has an essential role in the acquisition of natural re-
mechanisms have a role in relapse, as measured by re- ward and drug-seeking behavior. However, except in the
instatement of cocaine and heroin self-administration case of psychostimulants (see also Box 2), mesolimbic
after extinction. Selective dopamine-receptor antagonists dopamine neurotransmission does not seem to have a
attenuated reinstatement of heroin self-administration crucial role in reinforcement maintained by drugs of
induced by heroin priming injections57 but failed to abuse and natural rewards. Rather, it seems likely that

526 TINS Vol. 22, No. 11, 1999

 R. Spanagel and F. Weiss – Dopamine and reward REVIEW

dopaminergic-neuron activation highlights important 26 Tanda, G. et al. (1997) Eur. J. Neurosci. 9, 2077–2085
stimuli and functions as a learning signal. This function 27 Kalivas, P.W. and Stewart, J. (1991) Brain Res. Rev. 16, 223–244
28 Spanagel, R. (1995) Behav. Brain Res. 70, 37–49
would also account for the growing evidence of do- 29 Spanagel, R. and Hölter, S.M. in Handbook of Alcoholism (Zernig,
paminergic-neuron activation during reward expec- G., Kurz, M. and Saira, A., eds), CRC Press (in press)
tation, and increased dopamine release associated with 30 Rossetti, Z.L. et al. (1993) Eur. J. Pharmacol. 231, 203–207
31 Spanagel, R., Herz, A. and Shippenberg, T.S. (1990) J. Neurochem.
natural reinforcers under conditions of deprivation and 55, 1734–1740
novelty. These conclusions are in agreement with a re- 32 Johnson, S.W. and North, R.A. (1992) J. Neurosci. 12, 483–488
cent report demonstrating that, although activation of 33 Spanagel, R., Herz, A. and Shippenberg, T.S. (1992) Proc. Natl.
Acad. Sci. U. S. A. 89, 2046–2050
mesolimbic dopaminergic neurons is a necessary con-
34 Hemby, S.E. et al. (1995) J. Pharmacol. Exp. Ther. 273, 591–598
dition for intracranial self-stimulation, elicited dopamine 35 Ettenberg, A. et al. (1982) Psychopharmacology 78, 204–209
release is actually diminished during self-stimulation 36 Van Ree, J.M. and Ramsey, N. (1987) Eur. J. Pharmacol. 134,
of the mesolimbic system67. With regard to a role in 239–243
37 Pettit, H.O. et al. (1984) Psychopharmacology 84, 167–173
the development of compulsive drug-seeking behavior 38 Dworkin, S.I. et al. (1988) Pharmacol. Biochem. Behav. 30,
and dependence, neuroadaptive changes in midbrain 1051–1057
dopaminergic neurons might be involved in the gen- 39 Bozarth, M.A. (1987) Brain Res. 414, 77–83
40 Bals-Kubik, R. et al. (1993) J. Pharmacol. Exp. Ther. 264,
eration of deficit states that enhance vulnerability to 489–495
drug craving and relapse. 41 Shippenberg, T.S., Bals-Kubik, R. and Herz, A. (1993)
J. Pharmacol. Exp. Ther. 265, 53–59
Selected references 42 Maldonado, R. et al. (1997) Nature 388, 586–589
1 Olds, J. and Milner, P. (1954) J. Comp. Physiol. Psychol. 47, 419–426 43 Gessa, G.L. et al. (1985) Brain Res. 348, 201–203
2 Wise, R.A. and Rompre, P.P. (1989) Annu. Rev. Psychol. 40, 44 Imperato, A. and Di Chiara, G. (1986) J. Pharmacol. Exp. Ther.
191–225 239, 219–239
3 Di Chiara, G. and Imperato, A. (1988) Proc. Natl. Acad. Sci. 45 Kohl, R.R. et al. (1998) Psychopharmacology 139, 79–85
U. S. A. 85, 5274–5278 46 Gonzales, R.A. and Weiss, F. (1998) J. Neurosci. 18, 10663–10671
4 Pontieri, F.E., Tanda, G. and Di Chiara, G. (1995) Proc. Natl. 47 Rassnick, S., Stinus, L. and Koob, G.F. (1993) Brain Res. 623,
Acad. Sci. U. S. A. 92, 12304–12308 16–24
5 Koob, G.F. and Le Moal, M. (1997) Science 278, 52–58 48 Ikemoto, S. et al. (1997) Alcohol. Clin. Exp. Res. 21, 1042–1046
6 Wise, R.A. (1996) Curr. Opin. Neurobiol. 6, 243–251 49 Gatto, G.J. et al. (1994) Alcohol 11, 557–564
7 Robinson, T.E. and Berridge, K.C. (1993) Brain Res. Rev. 18, 50 Samson, H.H. et al. (1992) Ann. New York Acad. Sci. 654, 242–253
247–291 51 Hodge, C.W., Samson, H.H. and Chappelle, A.M. (1997) Alcohol.
8 Koob, G.F. (1992) Trends Pharmacol. Sci. 13, 177–184 Clin. Exp. Res. 21, 1083–1091
9 Spyraki, C. and Fibiger, H.C. (1988) Psychopharmacology 94, 52 Yoshimoto, K. et al. (1992) Alcohol. Clin. Exp. Res. 16, 781–785
133–137 53 Kiianmaa, K. et al. (1995) Pharmacol. Biochem. Behav. 52, 29–34
10 Balster, R.L. (1998) Drug Alcohol Depend. 51, 207–215 54 Nurmi, M. et al. (1996) Eur. J. Pharmacol. 315, 277–283
11 Young, A.M.J., Joseph, M.H. and Gray, J.A. (1993) Neuroscience 55 Weiss, F. et al. (1993) J. Pharmacol. Exp. Ther. 267, 250–258
54, 5–9 56 De Vries, T.J. et al. (1998) Eur. J. Neurosci. 10, 3565–3571
12 Wilson, C. et al. (1995) J. Neurosci. 15, 5169–5178 57 Shaham, Y. and Stewart, J. (1996) Psychopharmacology 125,
13 Bassareo, V. and Di Chiara, G. (1997) J. Neurosci. 17, 851–861 385–391
14 Mirenowicz, J. and Schultz, W. (1996) Nature 379, 449–451 58 Spanagel, R. and Zieglgänsberger, W. (1997) Trends Pharmacol.
15 Tanda, G. and Di Chiara, G. (1998) Eur. J. Neurosci. 10, 1179–1187 Sci. 18, 54–59
16 Schultz, W. (1998) J. Neurophysiol. 80, 1–27 59 Self, D.W. et al. (1996) Science 271, 1586–1589 Acknowledgements
17 Pettit, H.O. and Justice, J.B., Jr (1989) Pharmacol. Biochem. Behav. 60 Self, D.W. et al. (1998) J. Neurosci. 18, 1848–1859 The authors thank
34, 899–904 61 Nestby, P. et al. (1999) Psychopharmacology 141, 307–314 Walter
18 Di Ciano, P. et al. (1995) Behav. Pharmacol. 6, 311–322 62 Modell, J.G. et al. (1993) Alcohol. Clin. Exp. Res. 17, 234–240
63 Rossetti, Z.L., Hmaidan, Y. and Gessa, G.L. (1992) Eur. J. Zieglgänsberger for
19 Hemby, S.E. et al. (1997) Psychopharmacology 133, 7–16
20 Kiyatkin, E.A. and Stein, E.A. (1995) Neuroscience 64, 599–617 Pharmacol. 221, 227–234 generating the
21 Wise, R.A. et al. (1995) Psychopharmacology 120, 10–20 64 Diana, M. et al. (1996) Neuroscience 71, 411–415 figures. Their
22 Peoples, L.L. and West, M.O. (1996) J. Neurosci. 16, 3459–3473 65 Bailey, C.P. et al. (1998) Brain Res. 803, 144–152
research is supported
23 Roberts, D. et al. (1980) Pharmacol. Biochem. Behav. 12, 781–787 66 Kuhar, M.J. and Pilotte, N.S. (1996) Trends Pharmacol. Sci. 17,
24 Rocha, B.A. et al. (1998) Nat. Neurosci. 1, 132–137 260–264 by the BMBF
25 Sora, I. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 7699–7704 67 Garris, P.A. et al. (1999) Nature 398, 67–69 (01EB9706/3).

In the other Trends journals

Motion transparency: making models of motion perception transparent,
by Robert J. Snowden and Frans A.J. Verstraten Trends in Cognitive Sciences 3, 369–377
The neuroethology of primate vocal communication: substrates for the evolution of speech,
by Asif A. Ghazanfar and Marc D. Hauser Trends in Cognitive Sciences 3, 377–384
Emerging roles for RGS proteins in cell signalling, by J.R. Helper
Trends in Pharmacological Sciences 20, 376–382
GPCR–Ga fusion proteins: molecular analysis of receptor–G-protein coupling,
by R. Seifert, K. Wenzel-Siefert and B.K. Kobilka Trends in Pharmacological Sciences 20, 383–389
What do impact factors tell us?, by H. Kokko and W.J. Sutherland
Trends in Ecology and Evolution 14, 382–384
Protein translocation in apoptosis, by Alan G. Porter
Trends in Cell Biology 9, 394–401

TINS Vol. 22, No. 11, 1999 527