analogs produce substantial amounts of generalization to morphine,
amphetamine, pentobarbital, and nicotine, respectively. The fact
that there is less generalization across drug classes is an index of the
specificity of the drug stimulus. The cross-drug classifications which
have resulted from animal discrimination studies are generally
consistent with human data (Goldberg, Spealman, Shannon 1981).
For instance, if an animal has been trained to press one lever when
given amphetamine and another lever when given pentobarbital, it
tends to press the amphetamine lever more often than the pentobar-
bital lever following a nicotine injection (Schecter 1981). This finding
is consistent with that obtained in a study in which human
volunteers frequently identified nicotine injections as amphetamine
or cocaine at higher nicotine dose levels but not at the lower levels
and only rarely identified the nicotine injections as sedatives
(Henningfield, Miyasato, Jasinski 1985).

A more recent development is the extention of the systematic drug
discrimination procedures to use with human subjects. Similar
methods are used, and initial findings with drugs such as nicotine
and amphetamine are comparable to the results from animal studies
(Kallman et al. 1982; Chait, Uhlenhuth, Johanson 1984). Specifically
human volunteers can readily learn to differentially respond to the
presence or absence of these drugs, and the effects are dose related.

Drug Self-Administration

When given the mechanical means to do so, animals self-adminis-
ter addicting drugs (including nicotine) much like humans; that is,
drugs that function as rewards or reinforcers for humans also tend to
function as reinforcers for animals. The conceptualization of depen-
dence-producing drugs as reinforcers provided the framework for a
highly predictive test strategy, the self-administration study, where-
by animals or humans are given the opportunity to take drugs under
laboratory conditions (Thompson and Schuster 1968). This research
strategy permitted scientific analysis of the single common link
across all forms of drug dependence, namely that the addictive
behavior (for whatever reason) is motivated or controlled by the
drug’s reinforcing (rewarding) properties (Goldberg and Hoffmeister
1973; Thompson and Unna 1977; Seiden and Balster 1985). Stimuli
that can maintain and strengthen behavior leading to their presen-
tation are termed “positive reinforcers” regardless of their hypothe-
sized mechanism of action (e.g., alleviation of discomfort or produc-
tion of pleasure) (Skinner 1953; Thompson and Schuster 1968). The
reinforcing power or efficacy of a drug can be enhanced by a variety
of conditions (e.g., deprivation of the drug which the organism had
been repeatedly given, pain, food deprivation, social approval
contingent on drug taking, and perceived useful effects) (Thompson
and Schuster 1968; Thompson and Johanson 1981). Following

276
repeated exposure to a drug, a biologically mediated “drive” state
can be established that did not preexist as do the drives for food,
water, or sex.

The potential of a drug to serve as a reinforcer can be directly
assessed and quantified in laboratory studies of drug self-administra-
tion. Essentially, a human or animal subject is given access to the
drug; then his or her propensity to take the drug (i.e., to “self-
administer” the drug) can be measured. The self-administration test
provides the opportunity to rigorously study the main distinguishing
feature of drug dependence, that is, drug-seeking behavior. As is the
case in drug discrimination testing, animal data help to determine
the generality of the biological basis of the addictive process for a
given drug; for example, such data help to reveal if the process is
unique to humans because of social, genetic, or other factors. If the
drug is taken under a variety of prescribed conditions (summarized
later in this Section), then it is said to be functioning as a
“reinforcer” or “reward.”

The validity and generality of self-administration test results were
demonstrated by the observations that (1) there was a remarkable
degree of consistency between patterns of drug self-administration
among laboratory animals and observations concerning human drug
dependence (Jasinski 1977; Griffiths, Bigelow, Henningfield 1980),
(2) drugs that serve as reinforcers in self-administration studies also
tend to be “liked” when given to humans, and (3) there was a high
correlation among drugs which produced morphine-like euphoriant
effects and those which were self-administered by animals (Griffiths
and Balster 1979; Griffiths, Bigelow, Henningfield 1980; see related
data in Schuster, Fischman, Johanson 1981).

initiation of Drug Self-Administration

As discussed earlier in this Chapter, drugs cannot produce
dependence without initial exposure to them. Initiation of drug use
in humans is often mediated by social and other environmental
sources of pressure. To determine if a drug will reinforce behavior in
animals similarly requires some means of providing exposure to the
drug. Strategies for establishing drug taking in animals are analo-
gous in key respects to how humans may become dependent upon
drugs. Four general categories of methods are most commonly used.
The methods are not mutually exclusive and are sometimes used in
combination.

The first method of establishing drug self-administration in
animals is to provide initial doses (“priming” or “free sampling’’) and
then to gradually increase the dose (“graduation”). For instance, i.v.
drug infusions may be given to animals on a chronic basis while the
animals are also given the opportunities to take the drug. This
provides an opportunity to determine if simple exposure to the drug

277
is sufficient to result in drug seeking. A minor variation is to
gradually increase the dose of each injection over time. This general
procedure has been used to establish i.v. self-administration of d-
amphetamine, morphine, alcohol, pentobarbital, cocaine, nicotine,
and many other drugs (Deneau and Inoki 1967; Deneau, Yanagita,
Seevers 1969; Yanagita 1977; Woods, Ikomi, Winger 1971; Brady and
Lukas 1984; Griffiths, Bigelow, Henningfield 1980; Meisch 1987;
Henningfield and Goldberg 1983a).

A second method of establishing drug self-administration is to
substitute a new drug for one which was already serving as a
reinforcer. Humans do this as a function of drug availability; they
sometimes learn to like drugs which had not been taken previously
and may even come to prefer the new drug. Using this method with
animals provides a means of exposure to a new drug and may be
useful in comparing one drug with another. In animal studies,
cocaine is the most commonly used starter drug, because in animals
(as in humans) cocaine seems to be a source of reinforcement and/or
pleasure under an extremely broad range of conditions compared
with most other drugs. Variations on this procedure have been used
to evaluate the likelihood of self-administration of a wide range of
drugs including amphetamine, barbiturates, alcohol, opioids, and
nicotine (Griffiths et al. 1976, 1981; Woods 1980; Deneau 1977;
Yanagita 1977; Griffiths, Bigelow, Henningfield 1980; Brady and
Lukas 1984; Meisch 1987; Chapter ITD.

A third method is to induce the initial use of the test drug by
prearranged environmental sources of “pressure” or “motivation.”
Induction of drug taking can be accomplished with very explicit
contingencies. For example, presentation of food or withholding of
electric shock can be made contingent on drug consumption (Mello
and Mendelson 1971a,b). However, such direct contingencies often
result in minimal response output (i.e., drug consumption) to obtain
the positive reinforcer or to avoid the electric shock, and drug self-
administration may not persist after the contingencies are removed
(Mello 1973). For example, even when physical dependence on
alcohol had developed in rhesus monkeys, the animals often rejected
the drug when self-administration was not required to meet the
contingency (Mello and Mendelson 1971a). Thus, these procedures
have not been extensively used to generate animal models of human
drug taking (Griffiths, Bigelow, Henningfield 1980).

The fourth procedure for establishing drug self-administration
seems somewhat more analogous to how drug dependence may
sometimes develop in humans outside the laboratory, and has been
widely used to study drug self-administration in the laboratory; this
method is termed the “adjunctive behavior” or “schedule-induced
behavior” strategy (Falk 1983). The method involves a less direct
means of inducing drug intake; in fact, the drug does not need to be

278
taken to obtain the reinforcer or to avoid the punisher. Rather, the
animal is simply given the opportunity to take the drug; at the same
time, the experimenter arranges conditions that are highly likely to
engage the animal in cycles of work and breaking from work. For
example, the animal may have to press a lever to obtain food. The
result is that when the animal is unable to work on the food schedule
(e.g., during the brief “timeouts” or “waiting” periods), the animal
tends to take the drug. Eventually, the drug itself might come to
function as a reinforcer in its own right, even in the absence of the
environmental pressures that first led to its use. The dose level of the
drug is then increased gradually over time. Variations on this
procedure have been used to establish self-administration of alcohol
(Falk, Samson, Winger 1972; Freed, Carpenter, Hymowitz 1970;
Meisch 1975), pentobarbital (Meisch, Kliner, Henningfield 1981),
nicotine (Singer, Wallace, Hall 1982), and a variety of other drugs
(Brady and Lukas 1984; Meisch and Carroll 1981; Meisch 1987).
Although many environmental conditions are present outside the
laboratory that appear to function as do adjunctive schedules in the
establishment of human drug dependence (e.g., boredom in occupa-
tional settings), there have been few experimental studies of
adjunctive drug taking by humans (Falk 1983). One such study by
Cherek (1982) showed that volunteers took more puffs per cigarette
when they were given monetary reinforcers at regular intervals: the
volunteers had to press a button to obtain the reinforcer, but their
behavior did not decrease the time they had to wait for each
reinforcer to become available.

Evaluation of Reinforcing Effects

Conclusive demonstration that the effects of the drug itself were
the cause of the drug-seeking behavior is equivalent to showing that
the drug itself is functioning as a positive reinforcer. The basic
procedures were developed in animal studies (Pickens and Thompson
1968; Deneau 1977) and have been reviewed in detail elsewhere
(Johanson and Schuster 1981; Balster and Harris 1982; Fischman
and Schuster 1978; Yanagita 1980; Brady and Lukas 1984).

The most fundamental procedure is to verify that drug self-
administration occurs under conditions in which it is “optional” or
“voluntary”; that is, explicit contingencies for drug taking (e.g., to
obtain food, to avoid shock, or to obtain preferred liquid) are not
required. It is also necessary to ensure that the drug taking is not
simply maintained by the characteristics of the vehicle (e.g., water or
a flavored solution into which alcohol is placed, or the tobacco smoke
in which nicotine is delivered to smokers).

If the drug is serving as a reinforcing stimulus, it should be
capable of maintaining controlled behavior. For example, a complex
chain of drug seeking (i.e., “procurement”) might be required to

279
obtain the drug. An extension of this principle is to gradually
increase the amount of work (i.e., the ‘‘cost”) that must be expended
to achieve drug delivery to determine how much the subject works
(“pays”) for a given drug or drug dose. For example, the ratio of lever
press responses per drug injection is gradually increased in the
“Progressive Ratio” procedure to determine the maximum ratio
(“breaking point”) that will be sustained (Yanagita 1977; Griffiths,
Brady, Snell 1978a).

If the drug is serving as a reinforcer, then stimuli associated with
drug administration should also come to serve as reinforcers
(“conditioned reinforcers”). Of all dependence-producing drugs, the
importance of this factor may be most pronounced with regard to
nicotine because the various effects of nicotine may be associated
with tobacco smoke and other stimuli hundreds of times each day
over the course of many years of smoking. A fundamental observa-
tion is that even neutral-appearing stimuli can function as reinforc-
ers in their own right when they are associated (‘paired’) with
previously established reinforcers such as food, water, sex, or drugs
(Skinner 1953; Thompson and Schuster 1968). For example, the taste
and smell of alcohol are initially highly aversive to animals (Mello
1973), but in one study, the smell of alcohol was established as a
conditioned positive reinforcer for animals: the smell of alcohol was
enough to reinstate drug-seeking behavior even when the alcohol
was not physically available (Meisch 1977). Seemingly arbitrary
stimuli such as lights and tones can come to serve as reinforcers
after association with i.v. self-administered drugs including cocaine-
like stimulants, opioids, barbiturates, and nicotine (Goldberg 1970;
Goldberg, Kelleher, Morse 1975; Griffiths, Bigelow, Henningfield
1980; Goldberg et al. 1983).

The basic methods described above are also used in human drug
self-administration studies, although with various procedural adap-
tations which have been described in detail elsewhere (Nathan,
O’Brien, Lowenstein 1971; Cohen, Liebson, Faillace 1971; Mello,
McNamee, Mendelson 1968; Mello 1972; Meyer and Mirin 1979;
Bigelow, Griffiths, Liebson 1975; Henningfield, Lukas, Bigelow
1986). As in the animal drug self-administration studies, the human
volunteers must emit a measurable response that may lead to drug
ingestion: for example, riding an exercise bicycle (Griffiths, Bigelow,
Liebson 1979; Jones and Prada 1975) or pressing a button on a
portable work station (Mello and Mendelson 1978). Such work
requirements then become established as part of the chain of drug-
seeking behavior. They have an advantage over non-laboratory drug-
seeking behavior in that the amount of work can be carefully
measured. Such data provide quantitative estimates of the time
and/or work expended for drugs (see examples in the following
studies and reviews: Johanson and Uhlenhuth 1978; Bigelow,

280
Griffiths, Liebson 1975; Mello and Mendelson 1978; Fischman and
Schuster 1982; Henningfield and Goldberg 1983b; Jasinski, Johnson,
Henningfield 1984).

Results from Drug Self-Administration Studies

Most categories of drugs which have been found to cause wide-
spread drug dependence in the nonlaboratory setting have been
tested with animals and humans in laboratory settings. Results of
these studies have been reviewed in detail elsewhere (Griffiths,
Bigelow, Henningfield 1980; Brady and Lukas 1984; Henningfield,
Lukas, Bigelow 1986). Several categories of drugs have been found to
be self-administered by humans and animals in the laboratory
settings, to meet criteria as positive reinforcers, and to exhibit
orderly relations as a function of drug dose, drug pretreatment, and
other factors known to affect the intake of dependence-producing
drugs. These include alcohol, morphine, pentobarbital, amphet-
amine, cocaine, and nicotine in the forms of cigarettes and i.v.
injection.

Self-administration studies with animals are much more extensive
and have also been reviewed in detail elsewhere (Johanson and
Schuster 1981; Balster and Harris 1982; Fischman and Schuster
1978; Yanagita 1980; Brady and Lukas 1984; Young and Herling
1986). In brief, drug self-administration studies in animals in the
1960s showed that a range of drugs including opioids, amphetamines,
barbiturates, certain organic solvents, alcohol, cocaine, and nicotine
were self-administered (Weeks 1962; Thompson and Schuster 1964;
Deneau, Yanagita, Seevers 1969; Deneau and Inoki 1967). All of
these drugs were found to maintain powerful chains of drug-seeking
behavior, even when insufficient drug was taken to produce a
clinically significant degree of physical dependence (Goldberg,
Morse, Goldberg 1976). Drugs that did not serve as reinforcers in
these studies included caffeine, lysergic acid diethylamide (LSD), and
the major tranquilizer chlorpromazine.

The speed of drug delivery can affect its reinforcing efficacy (Kato,
Wakasa, Yanagita 1987). Thus, the inhaled form of cocaine (“crack”)
is considered more reinforcing and dependence producing than other
forms of cocaine delivery, with oral cocaine apparently among the
least reinforcing of the commonly used routes of delivery (see also
US DHHS 1987). Analogously, nicotine taken by the slow release
oral preparation (nicotine polacrilex gum) appears to be much less
reinforcing than nicotine taken by quicker release oral preparations
(e.g., chewing tobacco) or cigarette smoke (Chapters IV and VII).

Research findings have continued to extend the early observations
(Deneau, Yanagita, Seevers 1969) that the results with animals were
remarkably consistent with observations regarding human drug
dependence. For example, initial exposure of humans to drugs such

281
as opioids and stimulants led to addictive patterns of use, whereas
chlorpromazine rarely did, and LSD infrequently did (Jasinski 1977;
Griffiths et al. 1980). Earlier studies had suggested that alcohol,
caffeine, and nicotine were not reinforcers in animals (Mello 1973;
Russell 1979; Griffiths et al. 1986). However, by the early 1970s for
alcohol (Meisch and Thompson 1971; Meisch 1977, 1982) and 1981 for
nicotine (Goldberg, Spealman, Goldberg 1981), it had been confirmed
that these drugs could also serve as effective reinforcers for
nonhumans. The relatively little research done to assess the
dependence potential of caffeine has not as conclusively demon-
strated that it serves as a reinforcer in animals (Griffiths and
Woodson 1988b).

Drug Dose as a Determinant of Drug Intake

Drug dose per administration is a major factor that affects self-
administration of dependence-producing drugs. The resultant
dose-response relationships are orderly, and the data have been
reviewed extensively (Griffiths, Bigelow, Henningfield 1980; Johan-
son and Schuster 1981; Young and Herling 1986). In brief, the
relationship between the dose size available and the number of doses
taken is often referred to as an inverted U-shaped function because
of the shape of a graph that results when the number of injections (y-
axis) is plotted as a function of dose (x-axis) across a wide range of
doses to which a subject is given access.

Over the range of doses which appear to be functioning as effective
reinforcers, changes in dose are accompanied by compensatory
changes in number taken such that total drug intake is somewhat
stabilized. It appears that a determinant of such compensatory
changes in drug self-administration is the apparent upper and lower
“boundaries” or “thresholds” for aversive effects that might occur
when either too much drug is obtained or when insufficient drug is
obtained to prevent withdrawal responses (Kozlowski and Herman
1984). It should be noted, however, that in most studies, compensato-
ry changes in drug intake as dose level is changed are almost never
perfect and are frequently quite crude (Griffiths, Bigelow, Henning-
field 1980). (See Yokel and Pickens 1974 for an example of a study in
which drug intake was unusually stable across a range of amphet-
amine doses.) Thus, the usual observation related to drug dose is that
as dose is increased, the rate of drug taking decreases somewhat but
more total drug is obtained. This relationship is observed in studies
of i.v. nicotine in animals (Goldberg et al. 1983) and humans
(Henningfield, Miyasato, Jasinski 1983) and when tobacco smoke
dose is manipulated in humans (Chapter IV).

A misinterpretation of dose-response relationships by tobacco
researchers, largely in the 1970s, led to the controversy that marked
the so-called “titration studies” of tobacco intake. Specifically, it was

282
assumed that if a drug was serving as a reinforcer, then compensa-
tion for changes in dose level should have been more effective than
they appeared to be. Hence, some questioned whether nicotine was
serving as a reinforcer because dose-response relationships in
nicotine studies appeared very crude (Russell 1979). The question
that arose was not whether cigarette smokers showed compensatory
changes in responses to changes in dose level; they did. In fact, the
nicotine dose-response relationship has probably been better studied
and established, over a wider range of conditions and techniques of
study, than have dose-response relationships with any other class of
drugs which are self-administered by humans (Gritz 1980; Griffiths,
Bigelow, Henningfield 1980; Henningfield 1984). The question was,
rather, why compensatory changes in cigarette smoke intake often
_ appear to be inadequate to maintain stable levels of nicotine intake.
There are two main problems in interpreting these data, however.
The first is that in the vast majority of human cigarette smoking
studies, attempts to manipulate the dose delivered were not well
controlled and the measures used to assess the possible effects of
intended dose manipulations were not necessarily sensitive to
compensatory changes (see Chapter IV and Henningfield 1984b). The
second problem is that there is simply no basis for determining what
degree of compensation should occur, because the degree of compen-
sation observed in animal studies varies widely by drug and test
condition, and because there are relatively few human data involv-
ing drugs other than nicotine to which such a comparison might be
made (Griffiths, Bigelow, Henningfield 1980; Henningfield, Lukas,
Bigelow 1986).

Cost of the Drug as a Determinant of Intake

Cost of the drug is a determinant of intake in both laboratory and
non-laboratory settings. Evaluation of this phenomenon is objective-
ly carried out in the laboratory in which the amount of work
required to obtain the drug can be varied. From an economic
perspective, this is similar to varying the price of the commodity
which is available for purchase. Such manipulations with both
humans and animals have shown that cost (e.g., amount of work
required) affects drug intake: usually, the lower the cost, the greater
the intake. In some studies manipulations of both cost and drug dose
have been carried out (e.g., Moreton et al. 1977; Lemaire and Meisch
1985). These studies show that when the dose of the drug is reduced,
drug-seeking behavior may increase at first and thereby maintain
fairly stable intake, but if dose continues to decrease (or cost
continues to increase), the behavior will not be maintained (Lemaire
and Meish 1985). Early studies with cocaine, for example, showed
that if access to cocaine was limited, either by time or work (“cost”)
requirements, cocaine self-administration could be maintained indef-

283
initely without serious apparent adverse effects (Pickens and
Thompson 1968). However, if access to cocaine was nearly unlimited
and the cost requirement low, monkeys might self-administer toxic
dose levels (Deneau, Yanagita, Seevers 1969).

Use of tobacco in humans and intravenous nicotine self-adminis-
tration by animals appear to be similarly affected by manipulations
of cost as is use of other dependence-producing drugs. Specifically, as
the amount of work required to obtain nicotine injections in animals
is increased, the number of injections is decreased (Goldberg and
Henningfield, 1988). Analogously, human cigarette smokers and
other drug users can also be motivated with both positive and
negative cost incentives (Bigelow et al. 1981; McCaul et al. 1984;
Stitzer et al. 1982, 1986; Stitzer and Bigelow 1985). These laboratory
findings with animals and humans correspond to the effects of
changes in the price of cigarettes on cigarette sales (Lewit, Coate,
Grossman 1981; Lewit and Coate 1982; Warner 1986a). Such
relationships are also observed with other dependence-producing
drugs including opioids, sedatives, alcohol, and amphetamines
(Griffiths, Bigelow, Henningfield 1980; Yanagita 1977).

Place Conditioning Studies

Ingestion of dependence-producing drugs can lead to both positive
and negative associations with the setting in which the drug effects
were experienced. Whether the effects of a particular drug are
positive or negative depends on the dose that was given and other
factors that are discussed in this Section.

A scientific methodology for studying such phenomena is the
‘place-conditioning” or “place-preference-aversion” procedure (Bo-
zarth 1987a). This procedure provides an indirect means of assessing
the potential of a drug to establish drug seeking in the absence of
any explicit contingencies on the behavior. These procedures deter-
mine if exposure to a drug in a given environmental setting
enhances the preference of the animal for that setting. Conversely,
the procedure can be used to determine if exposure to a drug ina
specific environmental setting establishes an aversion of the animal
to that setting.

Because of their convenient size and the general validity of their
use as models for behavioral dependence potential testing, rats most
commonly are used as subjects in place-conditioning studies. The
general experimental procedure is to place the animal in one
environment (e.g., one chamber of a multiple-chamber test appara-
tus) when a drug is given and in another environment (e.g., distinct
in color, shape, or odor) when a placebo is given. Then, the animal is
given access to both environments (ie., placed in a connecting
passage or placed in one chamber or the other) to determine which
environment (chamber) it prefers (van der Kooy 1987; Bozarth

284
1987a), and, conversely, which environment it avoids. Studies have
shown that conditioned preferences can be established for morphine
(Bardo and Neisewander 1986), cocaine (Spyraki, Fibiger, Phillips
1982), alcohol (Stewart and Grupp 1985), and nicotine (Fudala, Teoh,
Iwamoto 1985; Fudala and Iwamoto 1987; Chapter IV).

The relevance of place conditioning as a factor that increases the
control of nicotine over behavior in human cigarette smokers may
exceed that of other dependence-producing drugs. This possibility
follows from the fact that the cigarette smoker has the ability to
readily produce a critical environmental cue associated with smok-
ing (cigarette smoke itself). Therefore, it should be possible for the
smoker to “enhance” the reinforcing efficacy of a range of environ-
ments (Iwamoto et al. 1987); the highly discriminating sight, smell,
and taste stimuli produced by tobacco smoke may effectively permit
the smoker to establish a “preferred environment.” This could
contribute to the dependence potential of nicotine. The observation
is also consistent with the finding that removal of the tobacco smoke-
associated stimuli is accompanied by decreased pleasure and/or
smoking (Gritz 1977; Goldfarb et al. 1976; Rose et al. 1987). As early
as 1899 it was observed, for example, “that the pleasure derived from
a pipe or cigar is abolished for many persons if the smoke is not seen,
as when it is smoked in the dark” (Cushny 1899).

Constraints on Dependence Potential Testing

The main constraint on procedures used to evaluate the depen-
dence potential of drugs is that they may fail to identify drugs which
only lead to dependence under unusual or uniquely human circum-
stances. For example, LSD does not serve as an effective reinforcer
for animals, and although its effects may be liked by humans under
certain conditions, it also produce feelings of fear, paranoia, and
other adverse effects (Griffiths, Bigelow, Henningfield 1980; Haert-
zen 1966, 1974). Caffeine provides an example of another kind of
drug which is sometimes used in the face of adverse effects, even
though the overwhelming majority of users do not use it in ways that
are considered to be of significant adverse health effect (Gilbert 1976;
Greden 1981). The anticholinergic drug atropine is another that is
representative of a class of drugs that occasionally are used in
nontherapeutic settings but do not appear to possess a marked
dependence potential when objectively tested (Penetar and Henning-
field 1986).

The wide range of factors that may result in occasional harmful
use of some substances (e.g., caffeine) or which may contribute to the
use of dependence-producing substances such as nicotine (Chapters
IV and VI) is not routinely explored in current laboratory depen-
dence potential tests. Thus, these drug dependence potential testing
procedures appear more likely to underestimate than to overesti-

285
mate the pharmacologic potential of a drug to cause dependence
outside of the laboratory. Furthermore, as discussed by Katz and
Goldberg (1988), because a variety of drug and nondrug factors
determine the actual prevalence of drug dependence outside of the
laboratory, dependence potential data are most reliable when
drawing qualitative conclusions. For example, such data are used to
determine whether a drug is dependence producing, or whether it is
more sedative- or stimulant-like.

Dependence Potential Testing: Tolerance and Withdrawal

In addition to taking control over behavior by virtue of reinforcing
and other behavior modifying effects, many addicting drugs can also
produce a physiological change termed physical dependence. Once
physically dependent, the person may experience an even greater
loss of control over use of a particular drug because abstinence from
the drug may be accompanied by discomfort and heightened urges to
take the drug (withdrawal syndrome).

Technically, physical dependence refers to physiological and
behavioral alterations that become increasingly manifest after
repeated exposure to a pharmacologic agent. As noted earlier, the
primary indication of physical dependence is the observation of drug-
abstinence-associated withdrawal signs and symptoms, although
tolerance is a frequent concomitant (Kalant 1978; Cochin 1970;
Kalant, LeBlanc, Gibbins 1971; Eddy 1973; Clouet and Iwatsubo
1975; Yanagita 1977). This phenomenon is also referred to as
“neuroadaptation” or “physiological” dependence (WHO 1981; Wool-
verton and Schuster 1983). It should be noted that use of the term
“physical” imports no greater degree of objectivity to phenomena
associated with physical dependence than to the phenomenon of
compulsive drug seeking: both physical dependence and drug seeking
involve physiologically mediated drug receptor interactions that
vary with the dose, kinetics, and type of drug. Furthermore, both of
these kinds of drug-associated phenomena involve behavioral and
physiological effects. For example, conventional measures of physi-
cal dependence include responses that are often considered behavior-
al (e.g., urge to use a drug, sleep time, food intake).

Research on opioid dependence in the 1940s focused largely on the
physical dependence that developed when opioids were given to
humans or certain animals (Martin and Isbell 1978). In particular,
characterizing the level of tolerance that was acquired when
morphine was repeatedly given, as well as the behavioral and
physiological sequelae of abrupt termination of such administration,
was a major contribution to the development of objective methods for
testing dependence-producing drugs in general. Observations emerg-
ing from such research in the 1940s led to strategies that are still
accepted as the definitive means to measure what may be termed the

286
TABLE 5.—Observations pertaining to the evaluation of
physical dependence potential, derived from
studies of morphine-like drugs

 

L Repeated drug administration leads to diminished responsiveness (i.e., tolerance) that is
more or less complete, depending upon the response measured. Responsiveness might be at
least partially overcome by increasing the dose. The degree of tolerance that develops is
generally directly related to the overall dosing level, but varies widely across various
possible measures.

2. The establishment of tolerance to one opioid is shared among many opium-derived and
related chemicals; the principle of “cross-tolerance” emerged as one means to further
classify a dependence-producing chemical.

3. Abrupt termination of use leads to behavioral and physiological responses that often tend to
be opposite of responses produced by acute drug administration. When these opposite
responses actually exceed normal baseline levels (e.g., opioid-induced constipation may be
replaced by diarrhea for a few days), they are termed “rebound” responses; hence the
frequent labeling of withdrawal as “rebound syndrome.” Together, these responses are
termed “the withdrawal syndrome.”

4. Severity of the withdrawal syndrome is related to the duration and dose levels of
preabstinence exposure to the drug.

ow

During withdrawal, readministration of the chronically given opioid can reverse the signs
and symptoms of the syndrome.

6. A range of opioids can substitute for the one to which an organism was chronically
exposed, thereby maintaining the level of physical dependence and preventing the onset of a
withdrawal reaction. These same drugs can be used to reverse the syndrome of withdrawal
precipitated by removal of the chronicaily given opioid. This observation provided the
rational basis for the systematic development of “substitution” or “replacement” therapy for
drug dependence.

 

NOTE: Details of the original experiments, and subsequent research upon which these observations follow, have
been reviewed (Martin and Isbel! 1978; Martin 1977; Sharp 1984; see also Deneau 1977).

“physical dependence potential” of a chemical (Jasinski 1977).
Specifically, these tests could be used to evaluate the likelihood that
(1) repeated use of a drug would lead to tolerance (physiological
adaptation) such that effects of repeated use would diminish and (2)
abrupt abstinence would be accompanied by a syndrome of behavior-
al and physiological disruption (withdrawal syndrome). Table 5
summarizes the prominent observations that emerged from these
early studies (Martin and Isbell 1978; Martin 1977). These observa-
tions provide the conceptual framework within which physical
dependence is assessed (Thompson and Unna 1977).

Tolerance

As noted earlier, repeated ingestion of most dependence-producing
drugs leads to diminished effects unless larger doses of the drug are
taken: this phenomenon is termed tolerance. One reason that
tolerance is an important factor in drug dependence is that it may
contribute to the escalation of drug self-administration that occurs
over time. This relationship is often misinterpreted, however.
Specifically, it is sometimes stated that tolerance results in a

287
continuous escalation of drug dose; however, lethal or aversive dose
levels prevent indefinite escalation.

Procedures for assessing tolerance development rely heavily on
procedures developed for assessing the direct effects of drugs
(Kalant, LeBlanc, Gibbins 1971; Abood 1984). Because psychoactive
drugs exert effects on numerous physiological systems and behavior-
al responses, almost any of a wide range of response measures can
serve in studies. Perhaps the most fundamental strategy of tolerance
assessment is to repeatedly present a given drug dose while
measuring the subsequent responses to drug administration. When
the response diminishes across drug presentations, tolerance to that
response is said to have occurred. Among the most frequent
measures of tolerance which have been used to assess psychoactive
drugs are discrimination of drug administration, analgesia, heart
rate, nausea, sedation, EEG activity, and performance on a behavior-
al task. Some measures (e.g., sedation from barbiturates) are more
specific to certain drug classes, whereas others (e.g., pleasurable and
dysphoric effects) are useful across a wider range of psychoactive
drugs. A variation on the foregoing procedure is to increase the drug
dose after responses have diminished to determine if the original
response level can be partially or completely restored.

Cross-Tolerance

Cross-tolerance is demonstrated when pretreatment with one drug
or formulation type produces tolerance to another drug or formula-
tion type (Wenger 1983; Yanura and Suzuki 1977; Martin and Fraser
1961). For example, a person who is maintained on an adequate dose
level of methadone will experience relatively little effect if he or she
injects his or her usual dose of heroin (Kreek 1979). Similarly,
persons given nicotine polacrilex gum may experience attenuated
effects from cigarettes, including reduced satisfaction from smoking
(Nemeth-Coslett et al. 1987).

Mechanisms of Tolerance

Several mechanisms of tolerance can be differentiated (Kalant,
LeBlanc, Gibbins 1971; Abood 1984; Haefely 1986; Sharp 1984; WHO
1981). For instance, if a drug impairs the ability to perform a task
that produces some form of reinforcement (e.g., humans working for
money or animals pressing a lever for food), the performance may
return to predrug exposure levels after repeated drug exposure over
time. In this example, at least four distinct mechanisms of tolerance
may have been operational; they are not mutually exclusive and may
co-occur (Kalant, LeBlanc, Gibbins 1971; Abood 1984; Haefely 1986;
Sharp 1984; WHO 1981; Eikelboom and Stewart 1979; Siegel 1975,
1976).

288
(1) The rate at which the drug was eliminated from the blood by
metabolism (detoxification) or excretion (in urine, feces, sweat, or
expired air) may have increased. This is frequently termed “disposi-
tional” or “metabolic” tolerance. A general method used to assess
dispositional tolerance is to measure the rate of decline in plasma
drug levels after varying amounts of drug exposure.

(2) The response at the cellular level might have decreased as the
drug receptor physiologically adapted to the drug or as the number
of receptors was altered (thereby functioning as though the systemic
dose had been reduced). This is frequently termed “functional” or
“pharmacodynamic” tolerance. One method used to assess function-
al tolerance is to hold the plasma drug levels constant while
measuring the response after varying amounts of drug exposure.

(3) The learning and motivational aspects of a behavioral situation
may have resulted in compensatory behaviors that reduced the
magnitude of the performance effects. This is frequently termed
“behavioral” tolerance, “drug sophistication,” or “behavioral adap-
tation.” Behavioral tolerance can be assessed by presenting the drug
at such long intervals so as to minimize the possible development of
functional or metabolic tolerance (e.g., Stitzer, Morrison, Domino
1970), or by using a variety of other controlled procedures (Krasne-
gor 1978b).

(4) Another behavioral mechanism that can lead to the develop-
ment of tolerance results from the classical or Pavlovian condition-
ing process that may occur where a drug is given. Pavlov (1927)
found that drug administration could produce an unconditioned
response that could subsequently occur as a conditioned response to
an associated environmental stimulus. However, sometimes the
conditioned response is opposite that of the drug response (Siegel
1975); when a drug-opposite response has been established, this
conditioning mechanism may reduce the strength of the response to
the drug itself (Goudie and Demellweek 1986).

The kinds of tolerance described above are sometimes categorized
together as “acquired” tolerance, which emphasizes the fact that
they have developed in an organism as a function of drug exposure
(WHO 1981). Tolerance development can be affected by the unit drug
dose, total daily dose, route of administration, prevailing environ-
mental stimuli, and exposure dynamics (exposure dynamics refers to
whether exposure to a drug is relatively continuous (Way, Loh, Shen
1969) or via multiple, discrete doses (Lukas, Moreton, Khazan 1982))
(see also, Dewey 1984; Adler and Geller 1984; O’Brien 1975; Blasig et
al. 1973; Okamoto, Rao, Walewski 1986). Acquired tolerance has
been demonstrated to occur with opioids and with most nonopioid
dependence-producing drugs, including nicotine (Martin 1977; Ka-
lant, LeBlanc, Gibbins 1971; Abood 1984; Haefely 1986; Domino
1973; Chapter III). In fact, classic techniques of measuring tolerance

289
evolved in a series of studies involving nicotine by Langley, Dixon,
and others near the end of the 19th century (Langley 1905; Dixon
and Lee 1912); these researchers found that tolerance to nicotine was
rapid and could be partially overcome by increasing the dose.

Constitutional Tolerance

Historically, although less commonly in recent years, tolerance
has been used to differentiate individuals or populations with regard
to their “preexisting” or “constitutional” level of drug responsive-
ness (Shuster 1984). This phenomenon has been designated “initial”
tolerance by a subcommittee of the WHO (WHO 1981) and is also
often referred to as “drug sensitivity” or “innate drug responsive-
ness.” The mechanisms may be similar to those described above; for
example, individuals may be born with differing numbers of
receptors for a particular drug or with different abilities to detoxify a
drug on the basis of enzymatic capacity of their liver. Analogously,
for reasons that are not related to drug exposure, certain populations
or individuals may be more effective in general at behaviorally
compensating for impediments to learning or performance. Genetic,
dietary, and early (including prenatal) developments are possible
sources of such variation that are under study (Abood 1984).

Whereas a fairly wide range of variation among such preexisting
levels of drug sensitivity has not been shown to affect the course of
development of drug dependence, extreme or qualitative differences
may have some impact. Such differences are sometimes held to alter
the vulnerability of various individuals or populations to the
development of drug addiction. One apparent example of such an
effect is the markedly higher percentage of Oriental persons who,
compared with most other populations in the United States, show an
aversive reaction to alcohol (‘‘flushing” response). This reaction
results from slower metabolism of the alcohol metabolite, acetylal-
dehyde, in Orientals compared with many other ethnic groupings
(Nagoshi et al. 1987). However, cultural factors also appear to
strongly influence rates of alcohol use in Orientals so that even
persons who show the flushing response may develop alcoholism
(Sue 1987; Johnson et al. 1987).

Differences in constitutional levels of tolerance among individuals
have been observed for all dependence-producing drugs, including
nicotine (Chapter II). However, the importance of such individual
and/or population differences remains unclear. In fact, a remarkable
feature of opioids, sedatives (including alcohol), and stimulants
(including nicotine) is the degree to which use has become en-
trenched in nearly any culture into which they have been introduced
(Austin 1979). Similarly, initial exposure to opioids, sedatives,
alcohol, cocaine-like stimulants, and nicotine has been shown for
each to lead to drug-seeking behavior in a wide range of animal

290
species including primates, dogs, and rodents (Deneau 1977; Yanagi-
ta 1977; Woods, Ikomi, Winger 1971; Brady and Lukas 1984;
Griffiths, Bigelow, Henningfield 1980; Meisch 1987; Meisch and
Carroll 1981).

Withdrawal Syndromes

As discussed earlier, documentation of a drug withdrawal syn-
drome is the primary line of evidence used to decide whether a
particular drug can cause physical dependence. The methods used to
properly conduct such tests and provide definitive results are
complex. This Section provides a summary of how such tests are
conducted and some of the main findings from tests of drugs such as
morphine, pentobarbital, and nicotine.

Measurement of drug withdrawal phenomena entails recording
physiological, subjective, and behavioral responses that occur when
drug administration is terminated, as well as those that occur
following drug administration. If the organism has developed a
sufficient degree of tolerance, such that levels of drug which
formerly disrupted physiological and behavioral functioning have
become necessary for relatively normal functioning, then the
organism is said to be physically dependent. Such drug abstinence-
induced disruption of functioning is termed a drug “withdrawal” or
“abstinence” reaction or syndrome. The behavioral and physiological
responses include some that are opposite those produced by drug
administration. For instance, opioid-induced pupillary constriction,
alcohol-induced muscle relaxation, and nicotine-induced tachycardia
may be replaced by pupillary dilation, convulsive muscle activity,
and bradycardia, respectively. Each drug withdrawal syndrome is
unique to a particular drug class and animal species and also varies
somewhat within individuals of a given species which are tested with
the same drug. Both frequency and magnitude of withdrawal
responses are typically measured.

In human studies, the range of measures available to assess
withdrawal reactions is considerable. They may be designated by
three categories: autonomic (e.g., blood pressure, pulse, core temper-
ature, respiratory rate, pupillary diameter, diarrhea), somatomotor
(e.g., nociception, neuromuscular reflexes, auditory and visual
evoked potentials), and behavioral (e.g., irritability, sleep/awake
cycle, hunger, urge to take the drug, ie., “craving”). Himmelsbach
and Andrews (1943) incorporated these distinctions into a weighted-
point system used for rating the severity of these signs and
symptoms of withdrawal (Fraser and Isbell 1960, Jasinski 1977).
Refinements in the scaling of opioid withdrawal responses have
continued (e.g., ARCI, weak opiate withdrawal scale) (Haertzen 1966;
Bradley et al. 1987; Handelsman et al. 1987).

291
Opioid withdrawal phenonena remain the most rigorously studied
and well characterized among the dependence-producing drugs. In
part, this is because of the ready observability of many of the signs
(e.g., dilated pupils, sweating, diarrhea). Other drugs for which
withdrawal reactions are now known or suspected to occur in
humans (e.g., amphetamine, cocaine, marijuana, phencyclidine) have
been much less thoroughly studied than the opioids and sedatives
(Mendelson and Mello 1984; Jones and Benowitz 1976). Studies with
these drugs are also hindered by the fact that there are fewer readily
observable signs of withdrawal, placing a greater burden on sophisti-
cated technology (e.g., EEG and neurohormonal assessment) and
procedures (e.g., performance assessment).

Two basic methods are used to measure withdrawal reactions.
After a period of chronic drug administration, behavioral and
physiological responses are measured following either abrupt drug
abstinence (“spontaneous withdrawal”) or the administration of a
drug antagonist (“precipitated withdrawal”) (Thompson and Unna
1977; Martin 1977).

Spontaneous Withdrawal Syndromes

Experimental studies of spontaneous withdrawal reactions include
two procedures for obtaining subjects which have been chronically
exposed to the drug. One procedure, termed the “direct addiction”
procedure, is to administer the drug to the subject at gradually
increasing dose levels, then to stabilize the dose for a predetermined
time interval. Drug administration is then abruptly discontinued,
and withdrawal measures are taken. This method has been used to
study withdrawal from opioids, barbiturates, benzodiazepines, stimu-
lants, ethanol, PCP, and gaseous anesthetics in a number of animal
species and humans (Brady and Lukas 1984). A variation on this
procedure is to abruptly withdraw subjects from a drug which they
had been chronically receiving in the nonlaboratory environment. In
human subjects, withdrawal reactions following cessation of use of
opioids, alcohol, nicotine, sedatives, and other drugs have been
studied using this procedure (Brady and Lukas 1984; Chapter IV).

A second procedure, termed the “substitution procedure,” involves
maintaining subjects at a given dose level of a standard or baseline
drug; periodically, doses of the standard drug are replaced with
either a placebo or a test drug to determine if there are signs of
withdrawal that occur before the next dose of the baseline drug
(Fraser 1957). This procedure provides information analogous to that
obtained from studies of cross-tolerance; namely, it permits determi-
nation of whether cross-dependence exists. If the test drug prevents
the expected onset of a withdrawal syndrome that should have
accompanied abstinence from the maintenance drug, then it is
possible that the two drugs produce similar kinds of physical

292
dependence. Because it is possible to suppress certain withdrawal
responses by using unrelated drugs (e.g., clonidine can suppress
certain aspects of morphine and nicotine (Jasinski, Johnson, Hen-
ningfield 1984)), a variety of control procedures are necessary to
identify the mechanism by which the replacement drug suppressed
the withdrawal responses (Martin 1977; Deneau and Weiss 1968;
Yanagita and Takahashi 1973; Okamoto, Rosenberg, Boisse 1975;
Jones, Prada, Martin 1976; Yanaura and Suzuki 1977).

In human subjects, both the direct addiction and substitution
strategies were used to evaluate withdrawal reactions from opioids,
barbiturates, and alcohol at the Addiction Research Center in the
1940s and 1950s (Himmelsbach 1941; Himmelsbach and Andrews
1943; Isbell et al. 1950, 1955). However, since those classic studies,
most dependence potential studies in humans have been conducted
with subjects who had been using the drug in a nonexperimental
setting prior to the study. The effects of abstinence from chronic
administration of opioids, barbiturates, benzodiazepines, caffeine,
and nicotine have been studied using these variations of spontaneous
withdrawal assessment (Benzer and Cushman 1980; Charney et al.
1981; Jaffe et al. 1983; Griffiths and Woodson 1988a; Greden 1981;
Hatsukami, Hughes, Pickens 1985; Chapter IV). A disadvantage of
such approaches is that it is not always possible to stabilize the
subjects at a known dose level, which results in considerable cross-
subject variation. The consequence of such dose-related variability is
that it can raise the threshold for the detection of significant effects.
This source of variability probably contributed to some of the earlier
inconsistent findings regarding the nature and severity of withdraw-
al reactions from tobacco (see further discussions in Murray and
Lawrence 1984). Early in the 20th century, analogous seemingly
inconsistent data led to debates about the existence of an alcohol
withdrawal syndrome (Isbell et al. 1955).

Precipitated Withdrawal Syndromes

Precipitated withdrawal responses may occur when a drug antago-
nist abruptly displaces the dependence-producing drug from its
binding sites on receptors. The viability of this approach depends on
the availability of a specific receptor antagonist which does not have
other actions that would preclude assessment of a withdrawal
syndrome. The antagonist is often given parenterally (e.g., intrave-
nously or intramuscularly) to maximize its rate of onset and hence
the likelihood of precipitating a withdrawal reaction.

Because of the availability of specific opioid antagonists, precipita-
tion of withdrawal phenomena associated with abstinence from the
morphine-like drugs has been most thoroughly studied using this
strategy (Martin et al. 1987). The studies have shown that the
process that leads to physical dependence begins with the first dose

293
of morphine (Higgins et al. 1987; Bickel et al. 1988) although such
low levels of physical dependence are not generally considered
sufficient for the clinical diagnosis of physical dependence. Analo-
gous studies have been conducted using the antagonists of the
benzodiazepines (e.g., diazepam (Lukas and Griffiths 1982, 1984)) and
are one element in the conclusive demonstration that these drugs do
produce physical dependence (WHO 1981, 1987). With regard to
tobacco or other forms of nicotine delivery, no such comparable
studies have been conducted, although, as discussed in Chapter IV,
preliminary and related data suggest the theoretical possibility that
nicotinic antagonists may be used to precipitate nicotine withdrawal
responses (Pickworth, Herning, Henningfield, 1988).

Variability in Withdrawal Syndromes

There are multiple determinants of the course and magnitude of
the withdrawal reaction from a drug. Factors which have been
studied in the laboratory are similar to those which affect the
development of tolerance described earlier. These include the total
daily dose of the drug that was given, specific drug type, the duration
of exposure, the schedule of termination, genetic constitution,
gender, and the prevailing environmental stimuli (Suzuki et al. 1987;
Suzuki et al. 1983; O’Brien et al. 1978; Suzuki et al. 1985; Yanagita
and Takahashi 1973; Yanagita 1973). In general, the magnitude of
the withdrawal reaction is directly correlated with the dose level
given, the duration of exposure, and the rapidity with which drug
levels at the receptor sites decrease. Conversely, lower dose levels,
shorter times of exposure, and gradual dose reduction (as opposed to
abrupt abstinence) can attenuate the withdrawal syndrome (Kalant,
LeBlanc, Gibbins 1971; Abood 1984; Jaffe 1985; Okamoto 1984).

Because withdrawal signs and symptoms vary among individuals
using the same drug, the syndrome may not be apparent when a
small number of individuals are studied. Lack of general understand-
ing of such factors probably contributed to the fact that the nature of
morphine withdrawal phenomena in humans was not rigorously
documented until the studies by Himmelsbach and his coworkers in
the 1940s (Himmelsbach 1941; Himmelsbach and Andrews 1943).
Similarly, withdrawal responses from chronic alcohol administra-
tion were not conclusively characterized and demonstrated until the
pioneering studies by Isbell and his coworkers in the 1950s (Isbell et
al. 1955). Research involving comparable strategies of assessment of
physical dependence on cocaine, amphetamine, marijuana, PCP, and
nicotine, only began in the late 1970s. In the absence of such data,
these drugs were sometimes held to be nonaddicting (e.g., President’s
Advisory Commission 1963). Nonetheless, for several of such drugs it
had long been recognized that some drug withdrawal phenomena did
occur (Jaffe 1970, 1976, 1980, 1985) and that such phenomena were

294
of clinical significance in the treatment of persons who were
attempting to abstain from them (Jaffe 1970, 1976, 1980, 1985,
Zweben 1986). For example, even prior to the rigorous studies of
tobacco withdrawal phenomena in the early 1980s (Chapter IV), the
Tobacco Withdrawal Syndrome had been recognized by the Ameri-
can Psychiatric Association (APA) as an Organic Mental Disorder in
its Diagnostic and Statistical Manual (DSM) of Mental Disorders
(APA 1980) on the basis of the extensive clinical observations and
other sources of information prior to the 1980s (Chapter IV). The
specificity of tobacco withdrawal to nicotine itself was acknowledged
in the revised DSM III (APA 1987).

Cravings or Urges

Among the most frequently discussed aspects of drug dependence
is the recurrent and often persistent urge to use drugs in drug-
dependent persons. The urge or desire to use a drug is widely termed
“craving.” However, how craving is defined and how craving-related
data are interpreted comprise one of the most problematic areas in
drug dependence research. For example, the term craving has been
used in such a variety of ways that its use may actually impede
accurate communication (Kozlowski and Wilkinson 1987; Henning-
field 1987). In the present Report, where possible, the term “craving”
has been replaced by more descriptive terms and phrases such as
“strength of an urge to use a drug” wherever the original meaning of
the referent material is not changed.

Whereas the urge to use a drug is a correlate of drug abstinence, it
is not an invariant one. For example, although urges to take drugs
reliably increase during early abstinence from morphine- and
pentobarbital-like (short-acting sedatives-hypnotics) drugs, they are
not a necessary concomitant of withdrawal reactions from other
opioids (e.g., cyclazocine) (Martin et al. 1965; Jasinski 1978), and
alcoholics often “voluntarily” abstain and undergo withdrawal even
when alcohol is available (Mello 1968; Mendelson and Mello 1966).
Moreover, such urges are also evoked by stimuli associated with
drugs and even by administration of the drug itself (O’Brien,
Ehrman, Ternes 1986; Childress et al., in press). Thus, urges to use
drugs also occur (often at high levels) when there is little other
evidence that physical dependence is present (e.g., many years after
drug abstinence) or when drug intake is sufficient so that no other
withdrawal signs or symptoms are present.

Because drug abstinence is only one of many factors that can
evoke the urge to use a drug and because such urges are not
necessarily alleviated by suppressing physiological withdrawal signs,
conclusions based upon such data must be carefully considered and
appropriately qualified. For instance, although methadone can block
withdrawal responses (at adequate dose levels), it does not reliably

295
diminish urges to use other opioids or opioid self-administration
(Jones and Prada 1975; Grabowski, Stitzer, Henningfield 1984;
Henningfield and Brown 1987). It would not be appropriate to
conclude that methadone did not effectively block withdrawal
reactions from morphine-like drugs simply because it did not
eliminate such urges, because by other measures, methadone is
effective at blocking opioid withdrawal (Kreek 1979; Jaffe 1985;
Jasinski and Henningfield 1988). Analogously. as reviewed in
Chapters IV and VII, most tobacco withdrawal responses are
effectively suppressed by nicotine replacement even though urges to
use cigarettes are not reliably diminished (see also Henningfield and
Jasinski 1988).

Constraints on Physical Dependence Potential Testing

There are both practical and conceptual constraints on physical
dependence potential testing. The practical constraints have been
discussed above and are related to the multiple sources of variability
in the intensity of withdrawal responses, which can result in failure
to detect withdrawal or in unreliable data.

The main conceptual constraint is that physical dependence is
neither a necessary nor sufficient condition to establish or maintain
drug-seeking behavior. For instance, drug-seeking and drug-taking
behaviors can persist at small doses of cocaine or morphine which
produce no significant degree of physical dependence in animals
(Schuster and Woods 1967; Deneau, Yanagita, Seevers 1969; Johan-
son, Balster, Bonese 1976; Jones and Prada 1977; Bozarth and Wise
1981) or in human subjects (Zinberg 1979). Conversely, animals in
the laboratory and humans in hospitals can be made physically
dependent on drugs such as opioids and barbiturates and yet never
display controlled or addictive drug-seeking behavior (WHO 1981;
Bell 1971). Similarly, compounds such as propranolol, cyclazocine,
and nitrites have clear physical dependence potentials in that
tolerance develops after repeated dosing and an abstinence syn-
drome appears upon cessation, yet drug-seeking or drug-taking
behavior does not reliably occur (Myers and Austin 1929; Crandall et
al. 1931; Rector, Seldon, Copenhaver 1955; Jasinski 1976; Jaffe 1985).

Another constraint is the difficulty in determining whether
abstinence-associated symptomology is specific to an individual or to
an underlying medical disorder that became evident upon removal of
the drug (Woody, McLellan, O’Brien 1984; Zweben 1986; Kosten,
Rounsaville, Kleber 1986; Stitzer and Gross 1988). For instance, an
opicid might alleviate depression in a person with primary affective
disorder. In general, as will be described below (s - “hapter IV),
withdrawal responses may be distinguished from J..e: . .bstinence-
associated symptomology by their relative consistency among indi-

296
viduals, by their transient nature, and by the direct relationship
between their magnitude and the level of preabstinence drug intake.

Finally, although the magnitude of the withdrawal syndrome is a
widely used index for assessing the degree of physical dependence, it
should be noted that this single measure is not always sufficient. For
instance, several studies have demonstrated that spontaneous with-
drawal from chronic levo-alpha-acetylmethadol (LAAM) or bupre-
norphine administration failed to result in pronounced signs of
withdrawal (Jasinski, Pevnick, Griffith 1978; Young, Steinfels,
Khazan 1979). Such observations could lead to the false conclusion
that LAAM and buprenorphine do not produce significant degrees of
physical dependence, when in fact a variety of other lines of evidence
confirm that they do. For example, administration of an opioid
antagonist such as naloxone precipitates a marked and intense
withdrawal syndrome in LAAM-maintained animals (Young, Stein-
fels, Khazan 1979). Analogously, Dum, Blasig, and Herz (1981)
performed a substitution type of experiment demonstrating that
chronic administration of buprenorphine also results in physical
dependence. The explanation for the misleadingly weak spontaneous
withdrawal phenomena for LAAM and buprenorphine seems to be
the slow elimination of these drugs from the plasma, which permits
the body to adjust more gradually to drug abstinence. The long
elimination half-life of LAAM’s active metabolites (Kaiko and
Inturrisi 1975) and buprenorphine’s unique affinity for the opiate
receptor and long elimination half-life (Cowan, Lewis, MacFarlane
1977) contribute to the lack of observed withdrawal signs after
chronic exposure is terminated. A similar example exists for the
long-acting benzodiazepine, diazepam. A delayed and relatively mild
withdrawal syndrome appears after spontaneous withdrawal, but
administration of the benzodiazepine receptor antagonist, Ro15-1788
(flumazenil), precipitates an immediate, intense abstinence syn-
drome (Lukas and Griffiths 1982, 1984). Analogous results are
produced when the daily dose level of shorter acting drugs is
gradually decreased.

A practical application of the finding that the magnitude of
withdrawal reactions tends to be inversely related to rate of drug
elimination is the gradual elimination of drugs from individuals who
are suspected of being highly physically dependent. Such gradual
elimination reduces the magnitude of the withdrawal syndrome.
This is the basis of the gradual withdrawal of morphine, alcohol, or
nicotine after a period of chronic intake at high dose levels (Jaffe
1985). Although gradual dose reduction of opioids and nicotine
reduces the magnitude of most aspects of the withdrawal syndrome,
it is not clear that such an approach improves overall treatment
outcome compared with much more rapid drug cessation (i.e., “cold
turkey”) (Jasinski and Henningfield 1988; Chapter VII).

297
Therapeutic or Useful Effects of Dependence-Producing
Drugs

With many dependence-producing drugs, the same biological
properties that are important in their dependence-producing proper-
ties may also lend them to therapeutic application. In fact, most
classes of drugs which cause dependence, including opioids, seda-
tives, alcohol, cocaine-like drugs, and nicotine, have been used as
medicinals to treat specific medical disorders and human discom-
forts. Descriptions of the approved and general uses are available in
the American Hospital Formulary Service (1988), the Physician’s
Desk Reference (Medical Economics Company 1988), the United
States Pharmacopeia (Griffiths, Fleeger, Miller 1986), and Goodman
and Gilman’s Pharmacological Basis of Therapeutics (Gilman et al.
1985) (see also Table 6).

Although each of the drugs listed in Table 6 has a range of
potential or actual therapeutic applications, past and current uses
are often related to their effects on mood, feeling, and behavior. For
instance, the stimulants may be used to modulate arousal level, the
opioids to alleviate pain, the sedatives to alleviate anxiety; the drugs
are sometimes systematically used to treat the dependence which
may have previously developed on them or on another drug in the
same class. Nicotine is no exception to these observations. Historical-
ly, tobacco was used to treat a range of disease states, although
usually without evidence of efficacy (Corti 1931; Austin 1979).
Nicotine in the polacrilex gum form is a drug approved by the FDA
for treatment of nicotine dependence (see Chapter VID.

The therapeutic effects of dependence-producing drugs not only
illustrate an important point of commonality among these drugs, but
these effects also may be important in the drug dependence process
itself. Such potential drug actions can be important in the initiation,
maintenance, and relapse to drug dependence. The dependence
process may have been precipitated by the therapeutic use (medical-
ly approved or self-initiated) of a drug. The dependence process may
be exacerbated by the real or perceived benefit of the drug to the
individual as such actions strengthen the reinforcing power of the
drug. The therapeutic actions of a drug may be associated with
relapse to drug use after many years of abstinence. These aspects of
dependence potential as they pertain to nicotine are discussed in
Chapter VI.

Adverse and Toxic Drug Effects

As discussed earlier, adverse drug effects are important clinical
features of drug dependence. These effects may be used as factors in
objective determinations of the overall liability associated with a
drug (Yanagita 1987; Griffiths et al. 1985). For instance, chronic
administration of sedatives or alcohol can produce intoxication and

298
662

TABLE 6.—Effects that may be produced by addicting drugs

 

Attribute

Nicotine *

Cocaine

Morphine-like

Alcohol

 

Discriminable interoceptive
(subjective) effects

Produce dose-related increases
in self-reported “liking” scores

Produce elevated response on MBG
(euphoria) scale of ARC inventory

Positive reinforcer in animal
drug self-administration studies

Positive reinforcer in human
drug self-administration studies

+
Henningfield and Goldberg
(1985), Morrison and
Stephenson (1969)

+
Henningfield, Miyasato,
Jasinski (1985)

+
Henningfield, Miyasato,
Jasinski (1985)

+
Goldberg, Spealman,
Goldberg (1981), Deneau
and Inoki (1967), Ando and
Yanagita (1981),
Henningfield and Goldberg
(1983a)

+
Henningfield, Miyasato,
Jasinski (1983)

+
Fischman et al. (1976)

+
Henningfield et al. (1987)

+
Fischman et al. (1976)

+
Pickens and Thompson
(1968), Deneau et al. (1969)

+
Fischman and Schuster
(1982)

+
Terry and Pellens (1970)

+
Martin and Fraser (1961)

+
Haertzen et al. (1963)

+
Headlee, Coppock, Nichols
(1955), Thompson and
Schuster (1964)

+
Jones and Prada (1975)

+
Carpenter (1962)

+?
Mello (1968)

+
Henningfield et al. (1984),
Stitzer et al. (1981)

+
Deneau et al. (1969),
Winger and Woods (1973)

aa
Bigelow et al. (1975), de
Wit et al. (1987)
00€

TABLE 6.—-Continued

 

Attribute

Nicotine *

Cocaine

Morphine-like

Alcohol

 

Place conditioning

Physical dependence develops such that

withdrawal accompanies
abrupt abstinence

Tolerance develops

Therapeutic use in treatment of

medical disorder

+
Fudala, Teoh, Iwamoto
(1985)

+
Hatsukami et al. (1984),
Hughes and Hatsukami
(1986)

+
Langley (1905), Domino
(1978), Marks, Burch,
Collins (1983), Jones,
Farrell, Herning (1978)

+!
AMA (1983), Gilman et al.
(1985), Medical Economics
Company (1987), and others

+
Spyraki, Fibiger, Phillips
(1982)

+?
Carroll and Lac (1987),
Jones (1984)

+
Tatum and Seevers (1929),
Downs and Eddy (1932),
Woolverton and Schuster
(1978), Wood and Emmett-
Oglesby (1987)

+?
AMA (1983), Gilman et al.
(1985), Medical Economics
Company (1987), and others

+
Bardo and Neisewander
(1986)

+
Light and Torrance (1929a),
Kolb and Himmelsbach
(1938), Himmelsbach (1941)

+
Light and Torrance (1929b)

43
AMA (1983), Gilman et al.
(1985), Medical Economics
Company (1987), and others

+-
Stewart and Grupp (1985)

+
Isbell et al. (1955)

+
Goldberg (1943)

+-!
AMA (1983), Gilman et al.
(1985), Medical Economics
Company (1987), and others