BEHAVIORAL PERSPECTIVES ON THE NEUROSCIENCE OF DRUG ADDICTION

JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR
2005, 84, 667–681
NUMBER 3 (NOVEMBER)
BEHAVIORAL PERSPECTIVES ON THE NEUROSCIENCE OF DRUG ADDICTION
GAIL WINGER, JAMES H. WOODS, CHAD M. GALUSKA, AND TAMMY WADE-GALUSKA
UNIVERSITY OF MICHIGAN
Neuroscientific approaches to drug addiction traditionally have been based on the premise that
addiction is a process that results from brain changes that in turn result from chronic administration of
drugs of abuse. An alternative approach views drug addiction as a behavioral disorder in which drugs
function as preeminent reinforcers. Although there is a fundamental discrepancy between these two
approaches, the emerging neuroscience of reinforcement and choice behavior eventually may shed
light on the brain mechanisms involved in excessive drug use. Behavioral scientists could assist in this
understanding by devoting more attention to the assessment of differences in the reinforcing strength
of drugs and by attempting to develop and validate behavioral models of addiction.
Key words: Drug addiction, neuroscience
_______________________________________________________________________________
Drug addiction continues to take a massive
pharmacologists look for clues to the etiology
toll in terms of economic loss and human
and control of drug abuse in the effects of
misery. For the purpose of this article, we
drugs on the behavior of humans or animals
define drug addiction as the final outcome of
under controlled experimental conditions.
a process that begins with occasional drug-
Neuroscience, because it searches for rela-
taking, and ends with consumption of exces-
tionships between brain function and behav-
sive amounts of drug to the detriment of
ior, is in an especially appropriate position to
society and the individual. Drug addiction is
study the neural correlates of the behavior of
a chronic, relapsing disorder that provides
drug abuse, and neuroscientists have contrib-
research and treatment challenges to scientists
uted a tremendous amount to our understand-
from widely ranging disciplines. Among these,
ing of the effects of drugs of abuse on the
geneticists and epidemiologists are particularly
brain and nervous system. This article will
intrigued by the fact that drug-taking behavior
address some of the neuroscience research on
exists on a continuum in humans: some
the problem of drug abuse, but will touch only
people engage in it to excess, most in
on limited aspects of what is a massive area of
moderation, and many not at all.
scientific inquiry.
Clearly, genetic differences and specific
The first section on brain reward circuitry is
societal-environmental conditions can play
included because this is the neuroanatomical
a role in the development of drug abuse.
basis for virtually all hypotheses and research
Psychiatrists and clinical psychologists attend
on the neuroscience of drug abuse. The next
more to the individual characteristics of the
section describes some recent research from
drug abuser, and consider how other cognitive
a few of the many neuroscientists who have
co-morbidities, such as anxiety or depression,
concentrated their efforts on drug addiction.
contribute to the development and mainte-
What is interesting and somewhat distressing
nance of drug abuse and addiction. Pharma-
for behavioral scientists is how little behavior is
cologists tend to focus on the drugs them-
regarded in much of this research.
selves, studying their mechanisms of action
Because drug addiction generally is con-
and attempting to develop potential drug
ceived as a process caused directly by chronic
antagonists that might be useful in the
administration of drugs of abuse, the investi-
treatment of drug abuse. And behavioral
gative effort is concentrated on a search for
changes in the morphology or molecular
Preparation of this review was supported by USPHS
biology in relevant parts of the brain as
grants AA 013713 and DA 015449.
a function of chronic drug administration.
Address correspondence to Gail Winger, University of
The neuroscience skills and techniques used
Michigan, Department of Pharmacology, 1302 MSRB III,
by these investigations are formidable, and
Ann Arbor, Michigan 48109-0632 (e-mail: gwinger@
interesting neuronal changes have been ob-
umich.edu).
doi: 10.1901/jeab.2005.101-04
served following chronic drug administration.
667

---

668
GAIL WINGER et al.
Even more technically fascinating is the
Finally, the perspectives offered in this
possibility of recreating these localized neuro-
paper are limited by both space and our time.
biological changes in the absence of drug
It would add useful knowledge to review what
administration—in effect producing an ad-
is known about the neuroanatomy and neuro-
dicted brain without giving the addicting
physiology of Pavlovian conditioning (e.g.,
substance. Animals so treated then can be
Cardinal & Everitt, 2004; LeDoux, 2000). Also
evaluated behaviorally, to determine whether
useful would be a discussion of several
they are, in fact, addicted. Unfortunately, the
psychological theories that have been posited
behavioral measures used in these types of
to explain drug addiction, many of which were
studies often are weak and not validated,
put forward by neuroscientists and based on
bringing the conclusions of much of this work
hypothesized or observed drug-induced brain
into question.
changes (e.g., Koob & Le Moal, 1997; Lub-
Our generally negative critique of much of
man, Yucel, & Pantelis, 2004; Robinson &
the most prominent work on the neuroscience
Berridge, 1993, 2000). However, our interest
of drug abuse is followed by a description of
in data rather than theories led us to exclude
a strictly behavioral approach to drug abuse.
this conjunction of drug addiction and neu-
This behavioral approach is presented as an
roscience.
additional avenue to be explored by neuro-
scientists and others investigating drug abuse.
NEUROANATOMY OF REWARD
We argue that drug addiction involves the
excessive choice of a drug over other environ-
In a simplified view, the brain can be
mental stimuli, perhaps because the drug is
thought of as the locus for the retrieval,
a more potent reinforcer relative to competing
processing, and storage of information trans-
reinforcers in the addict’s life. This section
mitted from the environment via neurons. The
serves two purposes: one is to present an
brain also activates motor, autonomic, and/or
option to the neuroscience approach and to
endocrine output in response to both the
suggest how addiction can be described more
current and historical environment. Quite a bit
appropriately; the other is to prepare for the
is known about the location of the specific
next section which returns to neuroscience,
sensory input and motor output areas of the
but in a more behavioral context.
brain, but less is understood about the brain
In this next section, we describe some
structures and processes involved in the
experiments that indicate that neuroscience
formation of constructs such as values, emo-
embedded in a more behavioral context
tions, and memories (see review by Glimcher,
potentially can identify the neurological cor-
2003). The current general concept of input
relates of behavioral constructs, such as re-
and output anatomy within the brain involves
inforcer strength. For example, investigations
parallel sensory and motor circuits that can be
of the brain correlates of choice eventually
described as cortical-striatal-pallidal-thalamic-
may clarify the regions in the brain and the
cortical (Heimer, 2003). These circuits carry
patterns of brain activity that are correlated
information from either the somatosensory
with preference for higher magnitude rein-
cortex or the motor cortex to related areas in
forcers. Although the work described in this
the putamen. Circuits run medially from the
section does not involve studies of drugs as
putamen to the globus pallidus or to the
reinforcers per se, it easily could be extended
substantia nigra, and the globus pallidus sends
to complement a behavioral model of addic-
projections to specific areas of the thalamus.
tion.
The thalamus closes the loop by returning
The behavioral approach to drug addiction
information to local areas of the cortex. The
presupposes that drugs are not qualitatively
key concept in these circuits is topographical
different from non-drug reinforcers. The last
organization. Different parts of each of these
section of this article describes recent neuro-
anatomical areas are activated depending on
scientific studies that assess the validity of this
the nature (e.g., visual, auditory, nociceptive)
assumption. For example, do the brain
and location (e.g., finger, arm, head) of the
changes that accompany reinforcement differ
information being received and conveyed.
if the stimulus is cocaine compared with other
The association areas of the brain also may
non-drug stimuli?
be connected in several circuits that run

---

NEUROSCIENCE OF DRUG ABUSE
669
somewhat parallel to the input and output
What clearly is missing is how the various
circuits. One such circuit, called the anterior
circuits are interconnected to integrate the
cingulate
circuit
(Alexander, DeLong,
&
sensory input with the motor output. This
Strick, 1986), incorporates much of what is
question stimulates much of the research on
thought to be involved in the brain’s motiva-
the neurocircuitry of reward. Although the
tional pathways (Kalivas, Churchill, & Klite-
parallel circuitry suggests that integration is
nick, 1993). A locus of particular importance
possible at many cortical and subcortical levels,
in this pathway is the ventral striatum which
the precise pathways whereby sensory input is
includes the nucleus accumbens. Dopamine
evaluated, modified by processes associated
release in this area has been considered
with memorial, emotional, and motivational
a critical mediator of the reinforcing effects
factors, and converted into motor output have
of stimuli, including drugs of abuse (Koob &
not been delineated clearly. It has been
LeMoal, 1997; Robbins & Everett, 1996),
thought that the nucleus accumbens, with its
although some investigators suggest a more
shell and core components, as well as its strong
general activating effect of dopamine (Horvitz,
connection to various amygdaloid nuclei, is
2000). Dopamine is released into the ventral
critical to emotional coordination of behavior-
striatum via dopaminergic neurons that pro-
al output.
ject from the ventral tegmentum (Figure 1).
The extended amygdala is a fairly recently
The ventral tegmentum also sends dopami-
described structure that is an important part of
nergic neurons to dorsal striatal and cortical
this system; it consists of two components. One
areas. The ventral striatum has non-dopami-
includes the central amygdaloid nucleus and
nergic, primarily glutamatergic, input from
the lateral bed nucleus of the stria terminalis,
the amygdala and hippocampus, which are
and carves a sweeping circuit beneath the
involved in memory and emotional processing.
main body of the striatum and the globus
Many of these pathways are reciprocal, with
pallidus. The other consists of the medial
information passing back to the structure that
amygdaloid nucleus and the medial bed
originated the stimulation.
nucleus of the stria terminalis, and travels
The cortical areas involved in reward cir-
along with the central component under the
cuitry are thought to include the entorhinal
lenticular nuclei (Alheid, 2003). The central
and perirhinal cortices, the anterior cingulate
and medial nuclei of the amygdala have
cortex, the temporal lobe and the posterior
substantial input from the basal lateral nucleus
area of the medial orbital frontal cortex.
of the amygdala and many projections to the
Projections from these cortical areas travel to
cortex and to autonomic and endocrine areas
the ventral striatum as well. In the ventral
in the hypothalamus and brain stem. As noted
striatum there is close interdigitation with
by Heimer (2003), ‘‘Thus, [the extended
pallidal structures, and there are neuronal
amygdala] represents a strategically placed
connections from the striatum to the ventral
ring formation capable of coordinating activ-
pallidum, the internal globus pallidus, and
ities in regions of the multiple limbic lobe
down to the substantia nigra pars reticulata.
forebrain for the development of coherent
The medial dorsal thalamus is innervated by
behavioral responses through the referenced
fibers from the pallidal areas, and thalamic
output channels’’ (p. 1735).
projections complete the circuit by traveling to
It is in the various parts of this reward circuit
the anterior cingulate cortex.
anatomy and its connection with motor
Thus, a cortical-striatal-pallidal-thalamic-cor-
circuitry where neuroscientists look for brain
tical circuit can be described in the emotional
changes that reflect reinforced behavior.
areas of the brain. The cortical areas are
different from those involved in motor control
NEUROSCIENTISTS LOOK FOR
and sensory input; the striatal and pallidal
DRUG-RELATED BRAIN CHANGES
areas typically are located more ventrally, have
considerable non-cortical input, and a distinct
Neuroscientists primarily interested in drug
thalamic nucleus is involved. Nevertheless, this
abuse typically search for changes in some
circuit has parallels with other primary circuits
aspect of the reward circuitry that accompany
in the brain (Alexander et al., 1986; Heimer,
the development of drug addiction. This
2003).
presupposes a consensus on what addiction

---

670
GAIL WINGER et al.
Fig. 1.
(Top) Structures related to reward pathway visible on midline of brain. Dark lines indicate dopamine
pathways. (Bottom) Relation of thalamus (with lateral thalamic nuclei) to the more lateral caudate nucleus and globus
pallidus.
is, and most neuroscientists agree that drug
disorder that persists as a sustained suscepti-
addiction is a behavioral disorder. They usually
bility to relapse (‘‘craving’’) even if an in-
define addiction in humans as a change over
dividual has abstained from drug use for an
time from occasional drug taking to more
extended period of time.
‘‘compulsive’’ drug taking, and a ‘‘loss of
Despite a common definition of the disor-
control’’ over the amount of drug taken.
der, and despite the fact that there are some
Addiction is further viewed as a chronic
models that might reflect compulsive drug

---

NEUROSCIENCE OF DRUG ABUSE
671
taking (Ahmed & Koob, 1998; Ahmed, Walker,
alterations in gene expression. Alterations in
& Koob, 2000; Deneau, Yanagita, & Seevers,
gene
expression,
with
their
consequent
1969), few of these neuroscientists use or
changes in protein synthesis, are considered
describe an animal model of ‘‘compulsive’’
to be a neural aspect of learning and memory,
or ‘‘uncontrolled’’ drug taking as part of their
or ‘‘plasticity’’ as it is referred to by neuro-
evaluation of accompanying brain changes.
scientists. Learning and memory may have
Measures of the reinforcing effects of drugs
neural mechanisms in common with drug
(drug self-administration) are used by several
abuse because both reflect changes in rein-
neuroscientists as indicators of the develop-
forced behavior over extended time periods.
ment of addiction. Even more popular are
Neuroscientists, stating that drug addiction is
measures of sensitization (an increase over time
a relatively stable condition, are especially
in the locomotor stimulating effects of a drug
interested in gene expression changes that
as a consequence of repeated administration
have an extended duration.
of the drug), conditioned place preference (a
There are many places where gene expres-
selection by the animal of an environment
sion can be modified. The group headed by
where it previously experienced the effect of
Eric Nestler has emphasized long-term, drug-
a drug), and reinstatement (an increase in
induced changes in transcription factors as
extinguished, drug-maintained responding as
a potentially critical aspect of gene expression
a consequence of non-contingent drug admin-
that might be related to drug addiction.
istration, stress, or other intervention). How-
Transcription factors are proteins that attach
ever, as indices of addiction, these measures
to the part of DNA that is responsible for the
leave much to be desired.
rate at which this DNA is transferred to RNA
Although drug self-administration proce-
and subsequently to the rate of synthesis of
dures seem to reflect the subjective drug
specific proteins. A number of neurotransmit-
effects reported by human drug addicts
ters act through their membrane receptors
(Griffiths & Balster, 1979), the relation be-
to
alter
levels
of
transcription
factors.
tween this and the critical outcome—excessive
Because drugs of abuse can directly affect the
behavior that is the hallmark of addiction—
action of some neurotransmitters on their
has not been determined.
receptors, this is a logical place to begin to
Sensitization has only the fact that it is
look for the effects of drugs on protein
produced
by
repeated
administration
of
synthesis.
a drug, and is maintained for several weeks
A recent review (Nestler, 2004) describes
or months following drug withdrawal, to
two transcription factors that his laboratory
recommend it as a model of addiction. This
has evaluated that meet some of their criteria
drug-induced increase in locomotion lacks
as important factors in drug addiction. Both of
even face validity as an indicator of compulsive
these, CREB (cAMP response element binding
drug taking or loss of control of drug-taking
protein) and deltaFos B, are increased in the
behavior. Conditioned place preference prob-
nucleus accumbens following chronic admin-
ably measures a different process from either
istration of cocaine. Elevated levels of CREB
reinforcement or sensitization (Bardo & Bev-
are fairly short lasting, declining within several
ins, 2000), but there is no evidence that this is
days after cocaine administration has been
related to addiction. Reinstatement has good
stopped. They also are associated with de-
face validity, but has not been shown to have
creased sensitivity to a wide variety of stimuli,
predictive validity as a model of relapse (Katz
including drugs of abuse (Barrot et al., 2002;
& Higgins, 2003).
McClung & Nestler, 2003), and seem unlikely
It can be argued that the lack of a validated
to be strongly related to drug addiction.
behavioral measure of addiction is the biggest
DeltaFos B, in contrast, increases a small
impediment for neuroscientists who focus on
amount with each injection of cocaine and
brain changes associated with chronic drug
is unusually stable in the brain, remaining
administration. Despite these challenges, the
at elevated levels for several weeks follow-
actual neuroscience is typically innovative and
ing termination of cocaine administration
productive. One of the most provocative
(McClung & Nestler, 2003). It is therefore
neural changes that occurs as a consequence
a more interesting candidate as a correlate of
of drug administration involves drug-induced
drug addiction.

---

672
GAIL WINGER et al.
Levels of deltaFos B can be modified in the
actions that oppose those of the D2-like
nucleus accumbens and the dorsal striatum of
receptors, and it is the D2-like receptors that
mice (the same areas in which they are
have been tied more closely to the stimulating
increased by cocaine administration) using
and addictive effects of cocaine.
technically sophisticated procedures that do
Research in the laboratory of Peter Kalivas
not involve cocaine administration. If elevated
has emphasized another protein, Homer, as
striatal deltaFos B levels play a role in drug
being particularly important in the develop-
addiction, mice with enhanced striatal delta-
ment of cocaine addiction. This research
Fos B levels in the absence of cocaine exposure
group has studied the importance of gluta-
should respond differently to cocaine than
mate, the excitatory neurotransmitter that is
mice with normal striatal deltaFos B levels. A
released by the neural connections that travel
variety of behavioral indices support this
from the prefrontal cortex and amygdala to
hypothesis. Mice with elevated striatal deltaFos
the nucleus accumbens and ventral tegmental
B levels exhibited greater preference for the
area, in mediating the neuronal plasticity
cocaine-paired compartment in a conditioned
involved in drug addiction. They suggest that
place preference study, relative to controls
dopamine release may underlie the develop-
(McClung & Nestler, 2003). In addition, these
ment of addiction, but that the permanent
mice self-administered smaller doses of co-
changes resulting in sensitization and rein-
caine than controls and completed higher
statement involve protein changes that regu-
ratios than control mice on progressive-ratio
late glutamate transmission (Kalivas, 2004).
schedules (Colby, Whisler, Steffen, Nestler, &
The Homer protein is associated with
Self, 2003).
glutamate receptors, where it acts as a scaffold
Interestingly, these effects appear to be
to keep relevant proteins in association with
somewhat specific to cocaine; the two groups
each other. Mice that had one of three Homer
did not differ in terms of food-maintained
genes knocked out developed a conditioned
responding. These results suggest that higher
place preference to a smaller dose of cocaine
striatal levels of deltaFos B may be correlated
than did normal mice and, at larger cocaine
with enhanced sensitivity to the incentive
doses, spent more time on the cocaine-
properties of cocaine. However, an important
associated area of the apparatus than did
comparison that was not made was to de-
normal mice. These knock-out mice also
termine whether mice with histories of chronic
showed a greater sensitization to cocaine than
cocaine self-administration, and presumably
did wild-type mice, and learned to self-admin-
cocaine-induced increases in deltaFos B, also
ister a small dose of cocaine more rapidly than
showed similar patterns of responding as those
wild-type littermates. In addition, the knock-
mice with artificially increased deltaFos B
out mice had reduced levels of accumbens
expression.
glutamate, similar to what had been found in
A driving concern of neuroscientists is
mice that had been given cocaine chronically
identification of the proteins that are synthe-
and then withdrawn. The similar effects of
sized in increased amounts as a result of
Homer deletion and cocaine withdrawal on
deltaFos B and CREB overexpression. Several
accumbens biochemistry, as well as the finding
candidates have been suggested, among them
that the Homer knock-out mice showed
a gut and brain protein, cholesystokinin, and
sensitization to the effects of cocaine without
an endogenous kappa receptor ligand, dynor-
having received prior injections of cocaine,
phin (McClung & Nestler, 2003; Nestler,
suggested that Homer proteins or the proteins
2004). A subunit of a glutamate receptor also
that are bonded together by Homer, are
has been found to increase with overexpres-
important in the development of cocaine
sion of deltaFos B (Nestler, 2004). In addition,
addiction (Szumlinski et al., 2004).
another peptide, Cdk5, is increased in animals
Terry Robinson has maintained an interest
that overexpress deltaFos B, and Cdk5 gene
in brain changes that accompany the develop-
expression also is increased by chronic cocaine
ment of sensitization to drugs of abuse. The
administration. Cdk5 may effectively reduce
incentive-sensitization theory developed by
the effects of dopamine that are produced
Robinson and Berridge (1993, 2000) posits
through the D1-like receptors (Benavides &
that the neural changes accompanying drug
Bibb, 2004). These receptors have some
administration are directly responsible for

---

NEUROSCIENCE OF DRUG ABUSE
673
both sensitization and the development of
when drug reinforcers assume control over
compulsive drug-taking behavior. Tests of this
a substantial portion of an individual’s behav-
theory have found that either the length of
ioral repertoire (Higgins, Heil, & Lussier,
neuronal dendrites, the number of branches
2004). As such, addiction to drugs can be
in the dendrites, and/or the concentration of
considered a form of excessive behavior,
the spines that come off the dendrites were
occurring when other activities are expected
increased in rats following either experiment-
and appropriate. Overeating and excessive
er-administered or self-administered amphet-
gambling are other examples of inappropriate
amine. Cocaine and nicotine produced similar
and excessive behaviors often attributed to an
increases in dendritic branching and density
addiction of some kind, but do not involve
in the accumbens and prefrontal cortex, but,
drug administration. A characteristic of each is
interestingly, morphine produced significant
that initial exposure to a reinforcing stimulus
decreases in these measures. The changes in
(e.g., euphoria, food, money) is followed by
dendritic morphology were relatively long
a progressive escalation in the behavior that
lasting, running in close parallel with the
produced it. Behavior that results in the
duration of sensitization (Robinson & Kolb,
availability of these reinforcers may eventually
2004), and the brain structures in which these
dominate the behavioral repertoire simply
changes were greatest differed depending on
because these stimuli function as more potent
the drugs (cocaine vs. amphetamine) given
reinforcers than others available in an individ-
and whether the drugs were given passively or
ual’s environment. This may be due, in part, to
taken by the animals.
genetic predispositions or, more likely, to
These studies of the molecular, biological,
particular learning histories combined with
and morphological changes that accompany
relatively easy access to these reinforcers (i.e.,
administration of drugs of abuse make the
a high rate of reinforcement) and insufficient
general point that such changes can, in fact,
contact with alternative sources of reinforce-
be measured. The neuroscientists who are
ment.
finding these drug-induced brain changes
One advantage of a behavioral approach to
have not claimed that these are necessarily
drug abuse is that, contrary to the drug-based
the essential components of drug addiction;
neuroscience theories, it not only accounts for
the brain is sufficiently complex and there are
excessive behavior that does not involve drugs,
so many aspects of its activity that can be
but it also accounts for situations in which
modified, that the search for correlates to
repeated exposure to drugs is not followed by
addiction is necessarily a long-term one. These
addiction. For example, people who use drugs
investigators, however, do hold to the debat-
to excess while they are young are likely to stop
able position that addiction constitutes drug-
using drugs when they get older, a process
induced changes in the brain.
called maturing out (Chen & Kandel, 1995).
We would like to propose an alternative to
When a young person is exposed to reinforcers
this drug-focused approach to drug addiction
that are incompatible with drug taking, such as
by treating drugs as reinforcing stimuli that
those associated with marriage, family, and
may come to dominate the behavioral reper-
employment, the relative reinforcing functions
toire of some individuals. This domination
of drugs usually decrease to the point where
may occur as the result of the relative efficacy
they no longer maintain the drug-taking
of certain drugs as reinforcers; that is, the
behavior. People who do not mature out of
strength of these reinforcers relative to other
their excessive drug taking may not have these
concurrently available reinforcers in the envi-
other reinforcers available, may not seek them
ronment. This approach is one we favor and is
out, or may not find them to be superior to the
described in more detail in the following
drugs they are taking due to particular
section.
learning histories and/or genetic predisposi-
tions.
As a second example of repeated drug use
DRUG ADDICTION AS A
BEHAVORAL DISORDER
not leading inevitably to addiction, consider
that soldiers who used heroin to excess while
Behavioral scientists generally regard drug
in combat situations in Vietnam typically did
addiction as a behavioral disorder that results
not continue this use when they returned

---

674
GAIL WINGER et al.
home (Robins, 1994). A third situation occurs
tists are beginning to look at activity in the
in patients who self-administer opioids for the
brain that accompanies choices between re-
treatment of pain but have no inclination to
inforcing stimuli that differ in magnitude or
continue to use the drug following recovery.
frequency, as described below (e.g., Cromwell
The reinforcing effect of the drug in this case
& Schultz, 2003; Platt & Glimcher, 1999).
is related to its ability to reduce pain, and
Although drugs have yet to be evaluated in this
following recovery there is no reason to
type of work, the work could easily progress to
continue to use the drug. There is also the
this point in the future, and understanding the
fact that a great many people have successfully
neurological correlates of the reinforcing
stopped smoking cigarettes, at least in part
strength of drugs may lead to a more unified
because the health risks became overwhelm-
account of drug addiction.
ingly obvious (Centers for Disease Control and
Prevention, 2004). It clearly is not the case that
ELECTROPHYSIOLOGICAL
simple exposure to drugs, even in the context
CORRELATES OF REINFORCERS
of their strong reinforcing effects, necessarily
OF DIFFERENT MAGNITUDE
leads to a permanent state of drug addiction.
A behavioral approach also is much more
Can neurophysiological and neurobiochem-
hopeful about the potential for treating drug
ical studies provide indicators of the relative
addiction (Higgins et al., 2004). Theories that
reinforcing effects of various stimuli? If so,
subscribe to drug-induced changes in the
knowing the neurological correlates of the
nervous system present more hopeless scenar-
relative reinforcing effects of stimuli might
ios
(once-a-drug-addict-always-a-drug-addict)
allow us to predict choice at the behavioral
that are more likely bereft of treatment
level. In addition, if addiction is accompanied
possibilities. Behavioral management of drug
by an increase in the reinforcing effects of
use, however, is one of the most successful
drugs relative to other stimuli, this could be
intervention strategies, particularly with co-
reflected in such neurological measures. Neu-
caine abuse for which there is no pharmaco-
ronal changes have been measured in various
logical treatment yet available. Contingency
parts of the brain upon the delivery of
management procedures typically involve giv-
a reinforcer. Experiments assessing brain
ing patients vouchers if they have drug-free
responses to reinforcers of different magni-
urine samples on a regular basis. The vouchers
tude often involve the use of a discrete-trial
can be exchanged for various goods and
delayed reaction task (Cromwell & Schultz,
services. Some contingency management ther-
2003; Hassani, Cromwell, & Schultz, 2001;
apies increase the value of the voucher over
Platt & Glimcher, 1999; Tremblay & Schultz,
time, as long as the client remains drug free,
1999; Watanabe, Hikosaka, Sakagami, & Shra-
and resets the value if cocaine use is detected
kawa, 2002). In a typical procedure, a monkey
or if the client refuses to submit a urine
learned to emit a response, either holding
sample. These procedures were far superior to
a lever or fixating on a visual stimulus, and to
standard therapy in producing drug-free cli-
maintain that response for several seconds.
ents and retaining them in treatment over a 24-
This resulted in the presentation of a discrim-
week study (Higgins et al., 1993). At this point,
inative stimulus (signaled delay) for a few
a behavioral approach is uniquely able to
seconds followed by the presentation of
generate successful strategies for prevention
another discriminative stimulus (trigger) that
and treatment.
signaled the availability of a reinforcer contin-
If one finds a behavior/reinforcement ap-
gent upon a manual or visual saccade re-
proach to drug abuse and addiction more
sponse. In some studies, it was necessary for
satisfying than the notion that drugs them-
the monkey to withhold responding following
selves are responsible for drug addiction, does
the presentation of one cue and to respond
that make neuroscience an irrelevant perspec-
following presentation of another cue (go-no
tive on drug addiction? Certainly not. But it
go task) either to earn a reinforcer or to
does mean that a rather different type of
advance to the next trial. In other studies,
neuroscience should be evaluated for what it
a discriminative stimulus indicated whether or
might contribute to our understanding of
not a reinforcer would be delivered upon the
addictive behavior. For example, neuroscien-
emission of the response. In a variation of the

---

NEUROSCIENCE OF DRUG ABUSE
675
task, the magnitude or type (preferred vs.
absence of reinforcement, but also which
nonpreferred) of reinforcer was signaled.
reinforcer would be absent. A small percent-
Neurons that fired during various aspects of
age of neurons fired according to a more
this task were recorded and described.
complex pattern: The rate of firing for some of
The activity of single neurons in a number
these neurons was highest during reinforced
of brain structures has been measured during
trials, prior to preferred reinforcers, and also
the delay between the presentation of the
during omissions of the less-preferred rein-
discriminative stimulus and the reinforced
forcer relative to omissions of the more
response. Neurons in the ventral tegmentum,
preferred reinforcer (Watanabe et al., 2002).
the ventral striatum, the caudate and putamen
The frequent finding that some neurons fired
(striatum), the orbital frontal cortex, and
at a higher rate prior to a signaled larger
various areas of the parietal and prefrontal
reinforcer and other neurons showed the
cortex have been studied most frequently. In
opposite pattern was a critical aspect of Gold
general, neurons located in all of the tested
and Shadlen’s (2001) description of brain
areas fired more rapidly during the delay when
decision-making based on sensory input.
a larger reinforcer, as opposed to a smaller
Behavioral measures in these types of studies
reinforcer, was signaled. Platt and Glimcher
also showed that responses to the larger or
(1999), using a visual saccade task, reported
more-preferred reinforcer differed from those
this differential response from neurons in the
to the smaller or less-preferred reinforcer
lateral intraparietal area of the posterior
(e.g., shorter latency or decreased probability
parietal cortex as well as in the superior
of visual fixation breaks; Watanabe et al.,
colliculus, a part of the visual pathway. In
2001). The possibility that these behavioral
several studies, a smaller proportion of neu-
differences were responsible for the different
rons were identified that fired in an exactly
rates of neuronal firing has been considered.
opposite pattern, showing greater activity
Although some research found reinforcer
when the smaller reinforcer was signaled.
magnitude-related differences in neuronal
Tremblay and Schultz (1999) found this
firing in the association cortex and none in
pattern of differentiation using recordings
the motor-related areas (Leon & Shadlen,
from the six-layer area of the orbitofrontal
1999), others observed that more precise
cortex of primates; a subset of orbitofrontal
recordings in the motor cortex clearly re-
cells fired more in response to cues predicting
flected the magnitude of the reinforcer
reinforcer A than to reinforcer B when A was
(Roesch & Olson, 2003). The latter investiga-
preferred, and more to B than to C when B was
tors subsequently reported that the orbito-
preferred. A similar pattern of differentiation
frontal cortex seemed to code the magnitude
was found in the anterior parts of caudate
of the reinforcer, firing most rapidly immedi-
nucleus,
putamen,
and
ventral
striatum
ately after presentation of the larger-reinforcer
(Cromwell & Schultz, 2003; Hassani et al.,
cue. Activity of the premotor cortex, in
2001).
contrast, was more related to motor prepara-
The lateral prefrontal cortex also appeared
tion, arousal, and attention; firing more
to code the significance of stimuli and re-
rapidly just before the trigger stimulus was
inforcement. Some neurons in this location
presented and when the reinforcer was de-
were particularly active during the delay that
livered (Roesch & Olson, 2004).
followed the onset of a stimulus correlated
These studies taken together indicate that
with reinforcement. Others were more active
several areas of the brain reward circuit
following the presentation of a stimulus cor-
contain neurons that are responsive to the
related with unreinforced trials. Some lateral
magnitude or value of signaled reinforcers.
prefrontal neurons fired more rapidly during
Various investigators have taken different
blocks of trials in which a preferred food (e.g.,
positions with respect to the significance of
apple rather than potato) or fluid maintained
these findings, some suggesting that these
responding. In addition, neurons that were
areas code reinforcer magnitude, others pro-
more responsive during unreinforced trials
posing that the differential firing rates reflect
also were more active when the less-preferred
consummatory or motor responses that are
reinforcer was available. These neurons, ac-
correlated with reinforcer magnitude. All seem
cording to the authors, coded not only the
to agree, however, that these simple firing

---

676
GAIL WINGER et al.
differences are not sufficient to explain the
reflect relative reinforcement rate immediately
formation of discriminations among reinforc-
when the visual stimuli were presented; the
ers of different magnitude or type, or to clarify
difference in firing rate as a function of local
how animals choose among reinforcers. None-
reinforcement rate increased over the time
theless, this research represents an important
course of the stimulus presentation and ‘‘re-
first step to a better understanding of where
set’’ after reinforcer delivery. The authors
and how decisions are made within the brain.
used this ‘‘reset effect’’ as one reason to
The techniques for single neuron recordings
speculate that the LIP is not the primary locus
eventually should permit simultaneous mea-
for computing reinforcement rate. The prima-
surements from a number of brain areas,
ry locus that encodes reinforcement rate
making it possible to track how incoming
should maintain that representation across
stimuli pass through the brain. Measurements
the appropriate time scale of the task. The
that follow the neuroanatomical description
LIP, the authors reasoned, may be involved in
of cortical-striatal-pallidal-thalamic-cortical cir-
the transformation of the color-based repre-
cuit eventually may lend physiological support
sentation of reinforcement rate to spatial eye-
to this notion, or demonstrate a much more
movement. This conforms to the notion of the
complicated interaction. Finally, theories such
LIP as an area that guides eye movements and
as those of Gold and Shadlen (2001, 2002),
shifts in visual attention based on incoming
who proposed mathematical rules of brain
information.
decision-making, will direct increasingly so-
Neuronal activity apparently related to the
phisticated research in this area.
outcome of a previous trial was observed in the
There have been two recent reports on
dorsolateral prefrontal cortex by Barraclough,
neuronal activity during more complicated
Conroy, and Lee (2004). Monkeys were
behavioral tasks involving choice. Sugrue,
trained to make saccade responses to one of
Corrado, and Newsome (2004) employed
two targets according to one of four different
a choice procedure in which the ratio of
strategies. In one, the two targets were
reinforcement for responding on two variable-
different colors, and a response to one of the
interval schedules varied several times over the
colors was reinforced 50% of the time. In the
course of a session. Monkeys modified their
other three algorithms, the computer engaged
selection of response options (visual saccades
the monkeys in a zero-sum game. In the first of
to one of two colored targets) quite rapidly
these three algorithms, the computer selected
within the session in accordance with the
the target to be reinforced on a random basis,
matching law (Herrnstein, 1961). To account
and the monkeys developed a side bias. In the
for the observed rapid sensitivity to local shifts
second, the computer took the animals’ pre-
in relative reinforcement rate, Sugrue et al.
vious choice behavior into account, and the
proposed a modification of the matching law
monkeys responded in a primarily win-stay-
that minimized the influence of previous
lose-switch manner. In the third, the computer
reinforcers through the use of a ‘‘leaky’’
took both choice and reinforcement history
integrator (for a competing model of dynamic
into account, and the monkeys responded in
choice, see Davison & Baum, 2000).
a fairly random manner. In each case, mon-
Sugrue et al. (2004) also recorded activity of
keys behaved optimally, with the optimal
single neurons in the lateral intraparietal
strategy resulting in a reinforcement probabil-
(LIP) area of the posterior parietal cortex, an
ity of 0.5.
area shown by Platt and Glimcher (1999) to be
Single neuron recordings taken from the
responsive to reinforcer magnitude when
dorsolateral prefrontal cortex during respond-
oculomotor decisions were required, and
ing maintained under the third algorithm
perhaps involved in spatial memory and motor
indicated that some neurons in this area were
planning. (e.g., Colby & Goldberg, 1999)
responsive to the choice and outcome of the
Interestingly, the authors reported that neu-
previous trial. For example, when a monkey
rons in this region exhibited a firing rate
selected a target, one neuron fired differen-
proportional to the relative reinforcement rate
tially based on whether the selection of that
of the stimulus associated with the alternative
target had been reinforced on the previous
currently presented within the cell’s response
trial; firing rates were higher if the selection
field. As a population, the neurons did not
on the previous trial was not reinforced.

---