Content uploaded by Stephen Tiffany
Author content
All content in this area was uploaded by Stephen Tiffany on Feb 06, 2018
Content may be subject to copyright.
Addiction (2000)95 (Supplement 2), S145–S153
HUMAN MODELS IN CRAVING RESEARCH
A cognitive processing model of alcohol
craving and compulsive alcohol use
STEPHEN T. TIFFANY & CYNTHIA A. CONKLIN
Department of Psychological Sciences, Purdue University, IN, USA
Abstract
Many addiction theories as sume that craving plays a central role in the acquisition and maintenance of drug
dependence. For example, craving is often depicted as the subjective exp erience of the motivational state
directly responsible for all drinking in the alcoholic. Craving has two prominent features that must be
explained by any viable model of craving. First, craving tends to be highly situationally speci c, readily
triggered by stimuli previously associated with drug use. Secondly, craving can persist well beyond the
cessation of drinking in an alcoholic. Conventional theories typically address craving’s cue speci city and
persistence by invoking concepts of classical conditioning. These theories fall into two classes: those that
emphasize withdrawal and those that focus on the positive-incentive properties of drugs. Both types of the ories
assume that craving processes are represented by the concomitant activation of craving report, drug-seeking
and drug use, and spec i c patterns of autonomic responses. How ever, research fails to nd more than modest
relationships across these putativ e manifestations of craving. The cognitive processing model, described in this
paper, offers a different view of craving’s form and function and proposes that drug use can operate
independently of the processes controlling craving. According to this model, addictive drug use is regulated by
automatic cognitive processes, while craving represents the activation of non-automatic processes. These
non-automatic processes are activated to either aid in completing interrupted drug use or block automatic
drug-use sequences. From this perspectiv e, craving is neither irrelevant nor central to the alcoholic’s drug use,
but rather serves as a cognitive marker of processes that, only in some instance, may be associated with alcohol
seeking and use. The research and treatment implications of this model’s assumptions regarding drug use and
craving processes are discussed.
Introduction
Craving is aversive, confusing, intrusive, frustrat-
ing, and exasperating. It can dominate your
thoughts and keep you awake at night. It can
compel irrational words and actions from what
might be otherwise reasonable people. For
many, craving represents the central dilemma of
their xation on addiction. Of course, all this
describes the impact of craving on many addic-
tions researchers; if craving is this vexatious for
the scientist, imagine its impact on the alcoholic.
Why is craving such a contentious issue in the
addictions eld? Craving may be so baf ing be-
cause it simply refuses to behave the way we
suppose it should. On one hand, we have rmly
entrenched assumptions regarding the role of
Correspondence to: Stephen T. Tiffany PhD, Department of Psychological Sciences, Purdue University, West
Lafayette, IN 47907, USA. Tel: (765)494 8509; fax: (765)496 2670; e-mail: tiffany@psych.purdue.edu
Submitted 1st September 1999; initial review completed 16th November 1999; nal version accepted 7th January
2000.
ISSN 0965–2140 print/ISSN 1360–0443 online/00/08S145–09 ÓSociety for the Study of Addiction to Alcohol and Other D rugs
Carfax Publishing, Taylor & Francis Ltd
S146 Stephen T. Tiffany & Cynthia A. Conklin
craving in addictive processes; on the other
hand, we have an accumulation of craving-
relevant data that consistently contradict our
most cherished suppositions about craving. What
are these assumptions and what are the contra-
dictory data? Furthermore, are there ways to
conceptualize craving so as to resolve these con-
tradictions? These are the questions to be ad-
dressed in the present paper.
Conventional co nceptualization of craving
Consider the following scenario: an alcoholic has
been discharged recently from an inpatient-
treatment program. He has not had a drink for
over a month and is fully con dent that he can
maintain his abstinence. Sure, there are times
when he craves a drink, but he’s able to control
his desire. One day, as he’s driving home after
work, he happens to pass by a bar he used to
frequent. Suddenly, his craving, which has been
so manageable, now seems overwhelming. Al-
most without thinking, he pulls into the parking
lot, enters the bar, orders a drink and gulps it
down. He has relapsed.
Common assumptions
The conventional explanation for the role of
craving in this scenario is relatively straightfor-
ward. It is commonly assumed that craving is at
the heart of all instances of alcohol use in the
alcoholic. That is, certain stimulus conditions,
for example the sight of the bar, trigger a state of
craving. Craving, in turn, generates compulsive
patterns of alcohol consumption. Embedded in
this model are the assumptions that craving op-
erates as a mediating variable that (1)is respon-
sible for the maintenance of all alcohol use in the
ongoing alcoholic and (2)provides the necessary
trigger for relapse in the alcoholic attempting to
remain sober.
The assumption that craving is the motiva-
tional core of alcoholism is deeply entrenched in
both scienti c and lay conceptualizations of al-
cohol dependence. The following statement by
Robinson & Berridge (1993)is representative of
the position held by many addiction scientists:
To understand addiction, therefore, we
need to understand the process by which
drug-taking behavior evolves into compulsive
drug-taking behavior. Presumably, this trans-
formation in behavior occurs because addicts
develop an obsessive craving for drugs, a crav-
ing that is so irresistible that it almost in-
evitably leads to drug seeking and drug taking
(p. 247).
This view of the central role played by craving
in addictive processes has several virtues, not the
least of which are the implications for treatment
development. For instance, Volpicelli and col-
leagues (Volpicelli et al., 1992)had this to say
about the development of medications for alco-
holism:
An ideal pharmacological adjunct to alco-
holism treatment should have the following
three properties: [with the rst being]it should
decrease the craving for alcohol so that the
motivation to drink decreases (p. 879).
Craving theories
Any tenable craving theory must explain, at a
minimum, two prominent features of craving.
First, craving episodes tend to be fairly situation
speci c. For instance, alcoholics will tell you that
situations or stimuli that they associate with pre-
vious alcohol use can be potent triggers of crav-
ing (e.g. Ludwig, 1986). Secondly, craving
episodes can occur months and even years after
cessation of drug use (Flaherty, McGuire &
Gatski, 1955; Mathew, Claghorn & Largen,
1979). Conventional theories of craving typically
address the situational speci city and persistence
of craving by invoking the concepts of classical
(Pavlovian)conditioning. In these theories, situ-
ations or cues that are routinely paired with drug
use are transformed into conditioned stimuli.
These stimuli produce conditioned responses
that motivate drug use.
There are two general classes of these types of
models: those that emphasize drug withdrawal
and those that emphasize a drug’s positive-incen-
tive properties. Withdrawal models hypothesize
that drug-paired stimuli elicit conditioned-with-
drawal effects (e.g. Ludwig & Wikler, 1974;
Poulos, Hinson & Siegel, 1981; Melchior &
Tabakoff, 1984). These effects, which are pre-
sumed to mimic the autonomic components of
drug withdrawal, generate drug craving. In the
case of our imaginary alcoholic, the sight of a bar
sign might serve as a conditioned stimulus that
activates components of the alcohol-withdrawal
syndrome. As a consequence, our alcoholic
Alcohol craving and c ompulsive alcohol use S147
would experience elements of autonomic hyper-
activity including sweating and a rapid pulse.
These conditioned withdrawal effects would trig-
ger craving that, in turn, would compel him to
seek out alcohol.
Positive-incentive models hypothesize that
drug-paired stimuli become conditioned incen-
tives that activate a central motivational state
(e.g. Stewart, de Wit & Eikelboom, 1984; Wise,
1988; Robinson & Berridge, 1993). This state,
which is conceptualized as craving, generates
craving report, drug-use behavior and autonomic
responses consistent with the direct effects of the
drug. According to this model, the bar sign was
an incentive stimulus that elicited craving and
approach behavior. In essence, this stimulus
acted as a magnet, drawing our alcoholic into the
bar and exposing him to even more incentive
stimuli. The net effect of this cascade of expo-
sures is that the alcoholic is pulled inexorably
into relapse. A major difference between this
type of model and the withdrawal models is that
the autonomic reactions elicited by conditioned
incentives are assumed to look like the direct,
activating effects of the drug, not the withdrawal-
like effects. Of course, in the case of alcohol this
is not a particularly useful distinction, as the
autonomic pro les of alcohol intoxication and
alcohol withdrawal can be somewhat similar
(Niaura et al., 1988).
Challenges to core assumptions
Withdrawal-based and incentive-based models
share several core assumptions. As noted above,
these theories assert that certain patterns of
physiological responses, particularly those con-
trolled by the autonomic nervous system, should
accompany the activation of craving states. Ac-
cording to the withdrawal models, withdrawal-
like physiology is necessary for the generation of
craving and, therefore, should co-vary strongly
with both self-report craving and drug-use mea-
sures. Similarly, incentive-based models gener-
ally assume that craving report, drug use and
autonomic responses should display at least a
moderate level of co-variation, as these responses
index the activation of the same drug-
motivational state. Furthermore, these models
adopt the standard assumption that self-report of
craving directly re ects the activation of the sub-
jective craving state and that craving is respon-
sible for all instances of drug use in the addict.
Therefore, craving report and drug use should be
tightly coupled. Such coupling should be evident
in laboratory-based studies of alcohol craving as
well as in investigations of the factors contribut-
ing to alcohol relapses.
There is scant support in either the alcoholism
literature or, more generally, across the drug-
abuse literature (Tiffany & Carter, 1998)for any
of the assumptions described above. Consider
the following.
Associations between craving report and cue-
speci c autonomic reactions. The procedure used
most commonly for studying craving in the labo-
ratory is the cue-reactivity paradigm (Drum-
mond et al., 1995). With this procedure, addicts
are presented with cues that have been, pre-
sumably, strongly associated with their previous
drug use. For example, alcoholics might be
asked to open a bottle of their favorite alcohol
beverage and sniff from the bottle. This research
evaluates the impact of these cues on alcoholics’
craving report, autonomic reactions and, in some
cases, drug-use behavior. We have examined 30
studies from the alcohol cue-reactivity literature
that had ratings of cue-speci c craving to drink
and some measure of autonomic reactivity. (The
list of the studies included in this analysis is
available from the authors.)These studies, which
had an average sample size of 27 alcoholics,
assessed a variety of autonomic variables includ-
ing heart rate, sweat-gland activity, blood press-
ure, peripheral temperature and salivation.
Across all these studies it would have been poss-
ible to calculate, theoretically, at least 73 corre-
lation coef cients between craving report and
autonomic measures. Only 14 of these possible
correlations were reported in this research; of
these, only three were signi cant. Thus, on bal-
ance, there is little evidence that cue-speci c
craving and cue-speci c autonomic reactions
display any systematic covariation.
Associations between drug use and cue-speci c
autonomic reactions. There are relatively few cue-
reactivity studies that permit an evaluation of
associations between cue-speci c autonomic re-
actions and some measure of alcohol choice or
alcohol consumption. We have identi ed eight
studies that might provide information abou t
these relationships. Of the 34 possible correla-
tions that could have been computed from these
S148 Stephen T. Tiffany & Cynthia A. Conklin
studies, 20 were reported. Of these, only three
were signi cant. Thus, there is no consistent
evidence of any systematic relationship between
measures of alcohol use and autonomic reac-
tivity.
Associations between craving and drug use: cue-
reactivity studies. These same studies also allow
us to examine relationships between craving re-
port and measures of alcohol use. This body of
research had the potential to yield at least 16
relevant coef cients of association, and 12 were
reported. Of those, only four were statistically
signi cant, and three of those four came from a
single study (Ludwig, Wikler & Stark, 1974).
Overall, laboratory-base d studies have provided
little evidence for the assumption that craving
and alcohol consumption are tightly coupled.
Associations between craving and drug use: relapse
studies. In the laboratory, craving and alcohol use
do not display the level of consistent, systematic
association mandated by the conventional as-
sumption that craving drives consumption. Is
there better support for this assumption from
studies of craving and relapse? If craving is at the
motivational core of alcoholism, then any inci-
dent of relapse should be preceded by a con-
spicuous episode of craving. Certainly, abstinent
alcoholics complain about the tribulations of
craving as they attempt to remain sober (Lud-
wig, 1988). However, relapsed alcoholics rarely
identify craving as a major precipitant of their
relapses. As one example, Miller & Gold (1994)
surveyed over 300 drug-dependent subjects
(mainly alcohol-dependent)who had partici-
pated in a drug abuse rehabilitation program 12
months prior to completing the survey. Those
who had not maintained continuous abstinence
since their discharge were asked to select a cause
for their rst relapse from a list of nine reasons
including craving. Fewer than 7% of the subjects
identi ed craving as a primary reason for their
relapse. Such results are fairly representative of
the relapse literature; alcoholics, heroin addicts,
cocaine addicts or cigarette smokers typically do
not describe craving as a pivotal factor in their
relapses (e.g. Marlatt & Gordon, 1980; Littman,
Stapleton & Oppenheim, 1983; Baer, Kamark &
Lichtenstein, 1989; Bradley et al., 1989; Ludwig,
1989; Wallace, 1989; Heather, Stallard & Teb-
butt, 1991).
Reactions to challenge
The collective message of the studies reviewed
above is not very supportive of the assumption
that craving is directly responsible for all or even
most instances of alcohol consumption in the
alcoholic. Addiction scientists have responded to
these challenges to craving orthodoxy in three
distinct ways. First, some researchers have dis-
counted the relevance of these unruly data. So,
for example, one could argue that assessments of
craving report tend to be unreliable or insensitive
to the genuine craving state of the addict. Such
crude measures may not be capable of revealing
true associations between craving and drug use
(e.g. Heather et al., 1991). Further, an argument
could be made that the sample sizes used in
these studies are too small to reveal any but the
most robust of correlations. Similarly, it may be
that a meta-analysis combining correlations
across studies would uncover signi cant associa-
tions that are not evident in single studies. In a
related vein, some have proposed that craving
operates at an unconscious level (e.g. Miller &
Gold, 1994; Berridge & Robinson, 1995). If
alcohol consumption is driven by unconscious
craving, there is no reason to assume that self-
reports of craving, which presumably are driven
by conscious processing, should be tightly cou-
pled to measures of consumption or autonomic
reactivity.
Secondly, it has been suggested that the con-
cept of craving could be banishe d from the ad-
dictions eld. According to this position,
arguments in favor of the utility of craving are
often circular and the concept, itself, is incapable
of complete reduction to operational de nitions
(e.g. Mello, 1972, 1978). From this perspective,
craving, as currently de ned, is at best an
epiphenomenon, and studies of the measure-
ment or manipulation of craving will not advance
the understanding or treatment of addictive dis-
orders.
Finally, it has been suggested that, although
craving may not operate as a necessary cause of
addictive drug use, craving may re ect the oper-
ation of processes important to any comprehen-
sive analysis of addictive disorders. For example,
the cognitive processing model of drug use and
craving (Tiffany, 1990)proposes that the regu-
lation of drug use in experienced addicts can
function independently of the processes that
control craving. According to this model, addic-
tive drug use is regulated by what cognitive
Alcohol craving and c ompulsive alcohol use S149
psychologists would describe as automatic pro-
cesses, whereas craving, in contrast, represents
the operation of non-automatic processes. These
non-automatic processes are activated, depend-
ing on the circumstances, to either aid or ob-
struct the execution of automatized sequences of
drug use. Therefore, according to this model,
craving is neither irrelevant nor central to the
control of drug use in the addict; instead, craving
may provide a marker of cognitive processes that
can, in some situations, moderate drug seeking
and drug use.
A cognitive processing model of alcohol use
and craving
From the time we wake in the morning to the
end of our day, we engage in a multitude of
activities controlled by cognitive processes that
operate quickly and effortlessly and require little
focused attention. Although it may seem easy to
groom, dress, cook, eat, walk, talk, read, drive,
work and play, none of these is a simple skill.
Many of our activities are so automatized, and
have been for so long that it may be dif cult to
remember what our performance was like when
we rst began to acquire these skills. However, at
the beginning, our execution of any of these
actions was hesitant, awkward and cognitively
challenging; it certainly required considerable ef-
fort, vigilance and focused attention. As we
practiced, our performance altered and what was
once demanding and clumsy became effortless
and highly coordinated. According to cognitive
psychologists, this transformation of perform-
ance with practice represents the transition from
non-automatic to automatic modes of function-
ing (Posner & Snyder, 1975; Shiffrin &
Schneider, 1977; Logan, 1988).
Major features of automatic processing. Automa-
tized performance can be characterized as having
the following features. First, with practice, au-
tomatized actions speed up and become less
variable. That is, automatized behaviors are in-
creasingly fast and stereotyped. Secondly, au-
tomatized actions tend to be stimulus bound.
The mere presentation of stimuli that have be-
come associated with the performance of an
action can trigger the execution of the automa-
tized sequence. Thirdly, in the presence of the
appropriate enabling stimuli, it may be dif cult
to inhibit an automatized action. Also, once ini-
tiated it may be exceedingly dif cult to prevent
the completion of an automatized sequence.
Thus, automatized actions exhibit a lack of con-
trol. Fourthly, automatized actions are easy and
non-demanding; they require relatively little cog-
nitive effort. Moreover, automatized actions gen-
erally do not interfere with other automatized
sequences or disrupt cognitive tasks that do re-
quire effort and attention. Finally, much of the
cognitive regulation of automatized performance
seems to occur outside of awareness.
Automatic processing and alcoholism. The cogni-
tive processing model proposes that, over a long
history of drinking, many of the actions required
to acquire and consume alcohol become automa-
tized for the alcoholic. That is, as with any other
skill, over the course of repeated practice, al-
cohol consumption in the alcoholic can be seen
as stimulus bound, stereotyped, effortless,
dif cult to control and largely regulated outside
of awareness. Many of the actions of the highly
practiced addict, both during regular use and
during relapse, may be viewed, not as a conse-
quence of craving, but as an example of the
kinds of behaviors exhibited by any automatized
skill (Tiffany & Carter, 1998). An excellent ex-
ample of automatic drinking was captured by
Ludwig (1988)in his consideration of the vari-
ous ways that alcoholics can experience compul-
sions to drink: “Others essentially think
instinctively, short circuiting both imagery and
cognitions, and are inclined to act without know-
ing why. When alcohol becomes readily avail-
able, they drink before they think” (p. 92).
The cognitive processing model acknowledges
that behavioral sequences as complex as alcohol
seeking and alcohol consumption can rarely be
regarded as being completely regulated by au-
tomatized functions. Skilled performance on
complex tasks is likely to be controlled by an
assembly of automatic and non-automatic pro-
cesses (Tiffany, 1990). However, to the extent
that components of those sequences are prac-
ticed repeatedly under consistent training condi-
tions, elements may become increasingly
automatized and integrated into higher-order au-
tomatized structures. Moreover, with repeated
practice of xed chains of behavior, even ele-
ments of behavior appearing early in the se-
quence may become increasingly regulated by
automatized processes.
S150 Stephen T. Tiffany & Cynthia A. Conklin
Major features of non-automatic processing. In
direct contrast to automatic functioning, non-
automatic processing can be characterized as
slow, exible, dependent upon intention, cogni-
tively effortful and restricted by limited cognitive
capacity. This kind of processing would be re-
quired under three kinds of circumstances. First,
as described above, the acquisition of any skill
typically requires, at the outset, some degree of
non-automatic processing. However, with prac-
tice, regulation of the skill is transferred to au-
tomatized modes of control. Secondly,
occasionally, even highly automatized sequences
encounter unexpected obstacles that cannot be
surmounted without the activation of non-
automatic forms of processing. For example, im-
agine that you are a highly practiced operator of
a television remote control. As you sit in front of
your television, mindlessly ipping through the
channels, the remote control suddenly dies. You
are now confronted with a problem that cannot
be resolved through those automatized se-
quences you use to operate the remote control.
This situation will demand the activation of non-
automatic processing to manage the problem.
Finally, it is widely accepted that it requires
considerable non-automatic processing to im-
pede or inhibit the execution of activated autom-
atized sequences (Shiffrin & Schneider, 1977).
This can be demonstrated fairly easily in the
following way: turn to any page in this paper and
try to count the number of periods and commas
without reading any of the text. Clearly, it will be
nearly impossible to keep from reading at least
some of the words you see. Reading is such a
highly automatized activity that it takes a great
deal of concentration and effort to prevent your-
self from doing it.
Non-automatic processing and craving. The cog-
nitive processing model hypothesizes that craving
represents the activation of non-automatic pro-
cesses that are activated in parallel with automa-
tized drug-use sequences. These are activated
either (1)in an attempt to overcome obstacles to
the completion of automatized drug use or (2)in
an attempt to prevent the execution of a mobi-
lized automatized sequence. These kinds of situ-
ations generate, respectively, the craving we see
in alcoholics who are not attempting to quit
using drugs and the craving associated with alco-
holics attempting abstinence.
Consider craving that comes about in alco-
holics who are not trying to remain sober but
who encounter some obstacle to their continued
drinking. What about the alcoholic who goes to
his local bar, the one where he does much of his
drinking, only to discover that it is closed for
remodeling? This obstacle will demand the acti-
vation of some measure of non-automatic cogni-
tive processing to solve this unexpected problem.
The non-automatic cognitive substrates of crav-
ing should be manifest in four domains. First,
overt behavior will be characterized by actions to
overcome the obstacles. For example, the al-
coholic may have to hunt for another bar or nd
a store and purchase alcohol. Secondly, self-re-
port should include expressions of desire for
alcohol and intentions to nd and consume al-
cohol. The alcoholic will also probably report
frustration and anger at being thwarted in his
attempt to drink. Thirdly, it is probable that the
alcoholic will display elements of autonomic acti-
vation that re ect either the physical or cognitive
demands of the craving situation. For example,
the actions required to secure a new source of
alcohol may require some physical exertion; if so,
autonomic systems will be mobilized to support
the metabolic demands of these actions. In ad-
dition, perhaps the solution to the alcoholic’s
dilemma is not immediately apparent. Conse-
quently, the cognitive demands of the situation
will also be associated with some degree of auto-
nomic activation.
These rst three kinds of variables (overt be-
havior, self-report and autonomic responses)
represent the categories of reactions traditionally
identi ed with craving activation. However, this
model proposes that craving processes may be
indexed by a fourth category of variables not
typically associated with craving assessment;
namely, indices of the cognitive demands of
craving. The cognitive-processing model posits
that craving represents the operation of capacity-
limited non-automatic cognitive processes.
These function at the cost of disruption of other
activities that also demand non-automatic pro-
cessing.
We have conducted research with cigarette
smokers on the cognitive demands of craving
that supports this prediction (Cepeda-Benito &
Tiffany, 1996). In this research, we induced
craving in the laboratory by havin g smokers viv-
idly imagine brief scenarios that describe desires
to smoke. Subjects reported that this manipu-
lation dramatically increased their craving rela-
Alcohol craving and c ompulsive alcohol use S151
tive to neutral-imagery conditions. During the
imagery manipulation, subjects were also asked
to press a button whenever they heard a brief
tone. Imagery-induced craving disrupted per-
formance in this simple task. That is, compared
to the neutral-imagery condition, smokers took
signi cantly longer to press the button during
craving imagery. Such demonstrations of the
cognitive demands of craving have been pro-
duced in other laboratories with both cigarette
(Sayette & Hufford, 1994)and alcohol craving
(Sayette et al., 1994). The results of this research
illustrate the exciting promise of a whole new
type of dependent variable for the study of crav-
ing processes.
Implications of the cognitive processing
model
The cognitive processing model presents an ap-
proach to craving distinctly different from the
traditional view of the form and function of
craving. Future research on alcohol craving and
treatment may wish to consider some of the
following issues suggested by this model:
·A core proposition of this cognitive model is
that alcohol use can operate independently of
the processes that control self-reports of crav-
ing. As reviewed in this paper, this proposal
certainly appears compatible with a consider-
able amount of research on alcohol craving
and alcohol use. If craving is not necessary for
alcohol consumption in the alcoholic, then
measures of self-reported craving cannot be
viewed as proxies for alcohol consumption or
indices of general alcohol motivation.
·Studies of autonomic responses generated in
studies of alcohol craving are nearly always
depicted as examples of classically conditioned
responses. According to the cognitive process-
ing model, many of the physiological changes
that accompany craving reports represent reac-
tions to the behavioral or cognitive demands of
the situation and are secondarily, if at all,
classically conditioned drug-withdrawal or
drug-appetitive effects. The possibility that the
autonomic physiology of craving is not limited
to classically conditioned responses allows
craving researchers to evaluate physiological
data from the broader perspective offered by
cognitive psychophysiolog y (e.g. Jennings &
Coles, 1991).
·The possibility that craving and consumption
are, at best, only loosely coupled, has pro-
found implications for the development of
medications for the treatment of alcoholism.
For example there may be no need to expect
that effective anticraving medications should
reduce alcohol consumption. Similarly, there
may be no need to assume that effective anti-
consumption medications should necessarily
reduce craving or, even if they do, that their
ef cacy derives from their impact on craving.
·The treatment approaches suggested by the
cognitive processing model focus not on the
alleviation of craving but, rather, on increasing
the chances that the alcoholic would be able to
successfully avoid or counteract automatized
drug-use routines. According to this model,
effective treatments should (a)target the stim-
uli that drive automatized drug-use routines,
or (b)work to protect or enhance the process-
ing resources required to impede the execution
of activated automatized sequences. With re-
gard to eliciting stimuli, any treatments that
eliminate or degrade the cues that trigger au-
tomatized drug-use sequences should decrease
the probably of alcohol use. For example, to
the extent that elements of alcohol withdrawal
reactions activate automatic alcohol-seeking
behaviors, a treatment that reduces withdrawal
should protect the alcoholic from relapse.
·Although treatments might be fashioned to
reduce an alcoholic’s exposure to evocative
cues, it is inconceivable that the abstinent
alcoholic could be protected from all stimuli
that might trigger automatic drug-use behav-
iors. Even under the best of circumstances,
maintaining abstinence in the face of such
stimuli would require considerable cognitive
effort. Non-automatic processing has limited
capacity, and anything that depletes process-
ing resources will decrease the chances that
the alcoholic will be able to successfully resist
automatized drug use. It is not likely, for in-
stance, that the alcoholic faced with the cogni-
tive challenges of coping with intense stress or
chronic negative affect would have the residual
resources to successfully battle automatized
drug-use behaviors. Any intervention that re-
duces ancillary demands on processing re-
sources should promote continued abstinence.
Similarly, any treatment that increases cogni-
tive capacity should aid abstinence. That is,
drugs with cognitive enhancing properties
S152 Stephen T. Tiffany & Cynthia A. Conklin
might provide useful adjuncts to a comprehen-
sive program of alcoholism treatment.
·Finally, most studies on the biology of craving
and alcohol use are interpreted within the
context of the putative motivational sig-
ni cance of craving. If craving is not at the
motivational core of alcoholism, then we may
wish to broaden our interpretation of such
studies. For example, over the next decade we
will see an increasing number of brain imaging
studies of the central nervous system (CNS)
correlates of cue-induced alcohol craving. The
traditional approach would most probably
portray any correlation between CNS acti-
vation and craving as a direct index of the
biological substrates of motivation to use al-
cohol. Although any such correlations may
re ect the biology of alcohol motivation, they
may also be revealing much about the biologi-
cal underpinnings of the cognitive organization
and processing demands of craving. Unless
those conducting brain-imaging research move
beyond a strictly motivational perspective on
craving, they may be unable to fully exploit the
information generated by this new technology.
References
BERRIDGE, K. C. & ROBIN SON, T . E. (199 5)The mind
of an addicted brain: neural sensitization of wanting
versus liking, Current Directions in Psychological
Science, 4, 71–76.
BAER, J. S., KAMARK, T., LICHTENSTEIN , E. & RANSOM,
C. C. (1989)Prediction of smoking relapse: analyses
of temptations and transgressions after initial cess-
ation, Journal of Consulting and Clinical Psychology,
57, 623–627.
BRADLEY, B. P., PHILLIP S, G., GREEN, L. & GOSS OP, M.
(1989)Circumstances surrounding the initial lapse
to opiate use following detoxi c ation, British Journal
of Psychiatry, 154 , 354–359.
CEPEDA-BENITO, A. & TIFFAN Y, S. T. (1996)The use
of a dual-task procedure for the assessment of
cognitive effort associated with smoking urges,
Psychopharmacology, 127, 155–163.
DRUM MOND, D. C., TI FFANY, S. T., GLAUTIER, S. P. &
REMINGTON, B. (Eds) (1995)Addictive Behaviour:
cue exposure, theory and practice (Chichester, John
Wiley & Sons).
FLAH ERTY, J. A., MCGUIRE, H . T. & GATSKI, R. L.
(1955)The psychodynamics of the “dry drunk”,
American Journal of Psychiatry, 112, 460–464.
HEATHER, N., STALLARD, A. & TEBBU TT, J. (199 1)Im-
portance of substance cues in relapse among heroin
users: comparison of two methods of investigation,
Addictive Behaviors, 16, 41–49.
JENNINGS, J. R. & COL ES, M. G. H. (Eds) (19 91)
Handbook of Cognitive Psychophysiology: central and
autonomic nervous system approaches (Chichester,
John Wiley & Sons).
LITTMAN, G. K., STAPLETON, J. & OPPENHEIM, A. N.
(1983)Situations related to alcoholism relapse,
British Journal of Addiction, 78, 381–389.
LOGAN, G. D. (1988)Toward an instance theory of
automatization, Psychological Review, 95, 492–527.
LUDWIG, A. M. (1986)Pavlov’s “bells” and alcohol
craving, Addictive Behaviors, 11, 87–91.
LUDWIG, A. M. (1988)Understanding the Alcoholic’s
Mind: the nature of craving and how to control it (New
York, Oxford University Press).
LUDWIG, A. (1989)The mystery of craving, Alcohol
Health Research World, 11, 12–17.
LUDWIG, A. M. & WIKLER, A. (1974)“Craving” and
relapse to drink, Quarterly Journal of Studies on
Alcohol, 35, 108–130.
LUDWIG, A. M., WIKLER, A. & STA RK, L. H. (1974)
The rst drink: psychobiological aspects of craving,
Archives of General Psychiatry, 30 , 539–547.
MARLATT, G. A. & GOR DON, J. R. (1 980)Determinants
of relapse: implications for the maintenance of be-
havior change, in: DAVID SON, P. O. & DAVIDSON, S.
M. (Eds)Behavioral Medicine: changing health
lifestyles, pp. 410–452 (New York, Brunne r/Mazel.
MATHEW, R. J., CLAGHORN, J. L. & LARGEN, J. (1979)
Craving for alcohol in sober alcoholics, American
Journal of Psychiatry, 136, 603–606.
MELCHIOR, C. L. & TABAKOFF , B. (1984)A condition-
ing model of alcohol tolerance, in: GALAN TER, M.
(Ed.)Recent Developments in Alcoholism: vol. 2,
pp. 5–16 (New York, Plenum Press).
MELLO, N. K. (1 972)Behavioral studies of alcoholism,
in: KISSIN, B. & BEGLEITER, H. (Eds)Biology of Alco-
holism, vol. 2, pp. 219–291 (New York, Plenum
Press).
MELLO, N. K. (1978)A se mantic aspect of alcoholism,
in: CAPPELL, H. D. & LEBLANC, A. E . (Eds)Biological
and Behavioral Approaches to Drug Dependence,
pp. 73–87 (Toronto, Addiction Research Foun-
dation).
MILLER, N. S. & GOLD, M. S. (1994)Dissociation of
“conscious desire” (craving)fr om and relapse in
alcohol and coc aine dependence, Annals of C linical
Psychiatry, 6, 99–106.
NIAURA, R. S., ROHSENOW, D. J., BINK OFF, J. A.,
MONTI, P . M., PEDRAZA, M. & ABRAMS, D. B. (1988)
Relevance of cue reactivity to understanding alcohol
and smoking relapse, Journal of Abnormal Psychology,
97, 133–152.
POSNER, M. I. & SNYDER, C. R. R. (1975)Attention
and cognitive control, in: R. L. SOL SO (Ed.)Infor-
mation Processing and C ognition: the Loyala sym po-
sium, pp. 55–85 (New York, John Wiley & Sons).
POULOS, C. X., HINSON, R. & SIEGEL, S. (1981)The
role of Pavlovian processes in drug tolerance and
dependence: implications for treatment, Addictive
Behaviors, 6, 205–211.
ROBINS ON, T. E. & BERRIDGE, K. C. (1993)The neural
basis of drug craving: an incentive-sensitization the-
ory of addiction, Brain Research Reviews, 18,
247–291.
Alcohol craving and c ompulsive alcohol use S153
SAYETTE, M. A. & HUFFORD , M. R. (1994)Effects of
cue-exposure and deprivation on cognitive re sources
in smokers, Journal of Abnormal Psychology, 103,
812–818.
SAYETTE, M. A., MONTI, P. M., ROSENHOW, D. J.,
GULLIVER, S. B., COLBY, S. M., SIROTA, A. D.,
NIAURA, R. & ABRAMS, D. A. (1994)The effects of
exposure on reaction time in male alcoholics, Journal
of Studies on Alcohol, 55, 629–633.
SHIFFR IN, R. M. & SCHNEIDER, W. (1977)Controlled
and automatic human information processing: II.
Perceptual learning, automati c attending, and a gen-
eral theory, Psychological Review, 84, 127–190.
STEWART, J., DEWIT , H. & EIKE LBOOM, R. (1984)Role
of unconditioned and conditioned drug effec ts in
self-administration of opiates and stimulants, P sycho-
logical Review, 91, 251–268.
TIFFAN Y, S. T. (19 90)A cognitive model of drug urges
and drug- use behavior: role of automatic and
nonautomatic processes, Psychological Review, 97,
147–168.
TIFFAN Y, S. T. & CARTER, B. (19 98)Is craving the
source of compulsive drug use? Journal of Psy-
chopharmacology, 12,23–30.
VOLPICELLI, J. R., ALTERMAN, A. I., HAYASHIDA, M. &
O’BRIEN, C. P. (1 992)Naltrexone in the treatment
of alcohol dependen ce, Archives of General
Psychiatry, 49, 8 76–880.
WALLACE, B. C. (1989)Psychological and environmen-
tal determinants of relapse in crack cocaine smokers,
Journal of Substance Abuse Treatment, 6, 95–106.
WISE, R. A. (1 988)The neurobiology of craving: impli-
cations for understanding and treatment of addic-
tion, Journal of Abnormal Psychology, 9 7, 118–132.