www.nature.

com/npp

REVIEW ARTICLE OPEN

Addiction is driven by excessive goal-directed drug choice

under negative affect: translational critique of habit and
compulsion theory
1
Lee Hogarth

Drug addiction may be a goal-directed choice driven by excessive drug value in negative affective states, a habit driven by strong
stimulus−response associations, or a compulsion driven by insensitivity to costs imposed on drug seeking. Laboratory animal and
human evidence for these three theories is evaluated. Excessive goal theory is supported by dependence severity being associated
with greater drug choice/economic demand. Drug choice is demonstrably goal-directed (driven by the expected value of the drug)
and can be augmented by stress/negative mood induction and withdrawal—effects amplified in those with psychiatric symptoms
and drug use coping motives. Furthermore, psychiatric symptoms confer risk of dependence, and coping motives mediate this risk.
Habit theory of addiction has weaker support. Habitual behaviour seen in drug-exposed animals often does not occur in complex
decision scenarios, or where responding is rewarded, so habit is unlikely to explain most human addictive behaviour where these
conditions apply. Furthermore, most human studies have not found greater propensity to habitual behaviour in drug users or as a
1234567890();,:

function of dependence severity, and the minority that have can be explained by task disengagement producing impaired explicit
contingency knowledge. Compulsion theory of addiction also has weak support. The persistence of punished drug seeking in
animals is better explained by greater drug value (evinced by the association with economic demand) than by insensitivity to costs.
Furthermore, human studies have provided weak evidence that propensity to discount cost imposed on drug seeking is associated
with dependence severity. These data suggest that human addiction is primarily driven by excessive goal-directed drug choice
under negative affect, and less by habit or compulsion. Addiction is pathological because negative states powerfully increase
expected drug value acutely outweighing abstinence goals.

Neuropsychopharmacology (2020) 0:1–16; https://doi.org/10.1038/s41386-020-0600-8

BRIEF INTRODUCTION TO THEORIES OF ADDICTION Positive reinforcement theories, by contrast, argue that the
Scientific and clinical theories seek to explain why addicts excessively rewarding effects of drugs drive persistent drug use
continue to take drugs despite experiencing consequential harms. independently of the withdrawal syndrome [12–15]. The
This paper evaluates evidence for just three accounts of addiction challenge for positive reinforcement theories is to explain why
—goal-directed choice under negative affect, habit and compul- drug use persists when addicts claim the drug has lost its value
sion—to try and determine which mechanism plays the most [16]. To solve this contradiction, a range of secondary processes
important role in addiction. A brief summary of the broad scope of have been postulated wherein drug seeking becomes less
addiction theory follows [1], to place these three accounts in controlled. Theories that appeal to appetitive Pavlovian con-
context. ditioning, for instance, argue that the pairing of drug cues with
Withdrawal-based negative reinforcement accounts argue that drug reward endows drug cues with capacity to elicit drug
although euphoric drug effects maintain initial use, growth in the seeking, and sensitivity to this effect may underpin dependence
adverse withdrawal syndrome drives persistent drug use [2–4]. [1, 17–20], because cue-reactivity is automatic [21], or because
The self-medication account specifies psychiatric symptoms, drug cues signal the accessibility (and hence greater utility) of
which increase during abstinence, as the main driver of the drug in the presence of drug cues [22, 23]. However, drug
persistent drug use [5]. Although individual sensitivity to a cue-reactivity is not reliably associated with dependence severity
multidimensional withdrawal syndrome and other negative in humans [24–27], suggesting this mechanism probably does
states is associated with problematic substance use, there not underpin addiction.
remains debate as to which component of withdrawal is most Other positive reinforcement accounts have argued drug-
important [6], how this relates to psychiatric comorbidity [7, 8], seeking behaviour becomes involuntary (or ‘crystalized’ [28]) with
and whether negative states prime drug seeking automatically practice. These theories argue that drug cues, or contexts, or the
[9] or via value-based decision making [10, 11]. This work forms completion of a prior link in a drug seeking chain, elicit drug
the basis of the goal-directed choice under negative affect seeking ‘automatically’ in the sense of not being determined by
account evaluated at length later. drug craving [29], or ‘habitually’ in the sense of not being

1
School of Psychology, University of Exeter, Washington Singer Building, Perry Road, Exeter EX4 4QG, UK
Correspondence: Lee Hogarth ([email protected])

Received: 5 July 2019 Revised: 9 December 2019 Accepted: 18 December 2019

© The Author(s) 2020

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
2
determined by an expectation of the current value of the drug
[30], or ‘compulsively’ in the sense that costs associated with drug Box 1: Definitions and methods of three addiction theories
Definitions
seeking are discounted and do not impinge on the behaviour
[31, 32]. The habit and compulsion models are evaluated at Theory 1—Goal-directed drug choice under negative affect
On this view, individuals differ in their experience of drug reward value, due to a
length later. variety of constitutional risk factors. The frequency of goal-directed drug seeking
Neurocognitive versions of positive reinforcement theory claim is driven by an expectation of drug value (determined by experienced value)
that although drug reward drives greater drug use, the persistence combined with knowledge of the voluntary behaviour necessary to obtain the
of this behaviour in addicted individuals is driven by acquired drug [8, 15, 297–299]. The second process is that for some users, negative states
(withdrawal, distress, pain, anxiety, depression etc.) are acutely mitigated by
dysfunction in decision-making capacity. These accounts differ in drugs, enabling these negative states to powerfully raise expected drug value
focus. Addiction could be driven by global impairments in (reflected in verbal ‘coping motives’), driving goal-directed drug seeking above
cognitive function [33], loss of volume/function of the prefrontal already elevated baselines [10, 11], acutely outweighing competing abstinence
cortex and other brain regions [34], specific impairments in goals [300].
inhibitory control [35], or specific narrowing of temporal horizon Theory 2—Habit.
Drug seeking is initially goal-directed, but drug reinforcement progressively
such that future costs and benefits are not considered in decision strengthens the association between drug stimuli (S) and the drug-seeking
making [36]. It remains unclear to what extent these neurocog- response (R), such that drug stimuli can elicit the drug-seeking response directly
nitive dysfunctions can be methodologically isolated from each through an S−R association, without retrieving an expectation of drug value [30].
other, and whether they play a causal/prospective role in Addiction is driven by an increased contribution of this S−R/reinforcement
addiction or are non-functional consequences of drug exposure mechanism to the control of drug seeking. Habit theory predicts that drug
seeking is controlled by the established S−R strength only so long as the
[37]. These neurocognitive models will not be considered further reinforcer is not re-experienced has having a different value. Once the reinforcer
because there is insufficient space to do justice to this broad field. is experienced as having a lower or higher value, this will change the strength of
The various theories for addiction are usually pitted against one the S−R association and the frequency of drug seeking will change accordingly
other on the assumption that only one mechanism can explain (i.e. habitual drug seeking is flexible, but requires experience of the changed
value of the drug to adapt).
addiction. However, multiple mechanisms could contribute
Theory 3—Compulsion
simultaneously. Furthermore, the underpinning mechanisms Compulsion theory is akin to habit theory, except that the flexibility of drug seeking
could differ between individuals depending on developmental is argued to be lost. As with habit theory, drug seeking is argued to be goal-directed
pathway, constitution, risk and protective factors [38, 39], or initially, and then transitions to become an S−R habit, but then in the third stage,
between drug classes, for example, stimulants vs. depressants [40], becomes a compulsion—a maladaptive habit where the S−R association
controlling drug seeking can no longer be modified by direct experience of the
or across types of behaviour, for example, drug seeking vs. drug drug reinforcer [31, 32]. Because compulsive drug seeking is controlled by the
taking [41, 42]. However, methods do not exist to adequately established S−R association (and not by the S−R/reinforcement mechanism) drug
isolate the contribution of specific mechanisms to behaviour in seeking is not modified by direct experience of the low value of the reinforcer and
different conditions. so continues in perpetuity despite loss of value. This explains why drug seeking
persists even though drug use is directly experienced as harmful, because this does
Box 1 defines the three theories of addiction evaluated in this not weaken the S−R association controlling drug seeking.
paper, alongside key methods used to test each theory in
laboratory animals and humans—providing a translational per- Methods
Stress-induced reinstatement in laboratory animals: In the reinstatement model,
spective. The methods used with each species are descriptively laboratory animals are first trained to self-administer an addictive drug. The
similar and arguably tap the same theoretical mechanisms, but response is then extinguished by omitting the drug, and responding declines. In
multiple methodological differences make direct comparison the reinstatement test, animals are exposed to stress vs. no-stress (manipulated
between species challenging. Despite these complications, the behaviourally or pharmacologically), and it is typically found that stress increases
(reinstates) drug self-administration [83].
weight of evidence does appear to provide converging transla-
Mood/stress-induced drug motivation in humans: Human drug users are exposed
tional support for the claim that addiction is primarily driven by to negative mood or stress induction via a range of methods (music, self-
excessive goal-directed drug choice under negative affect, and to referential statements, mood congruent words, public speaking, cold pressor,
a lesser degree by habit or compulsion. heat pain, serial addition, video clips, guided imagery etc.), and contrasted to a
no-induction control condition (either between-subjects or counterbalanced
within-subjects). Drug motivation is measured post-induction via a range of
methods (craving, choice, consumption, cognitive bias, economic demand etc.). It
IS ADDICTION PRIMARILY DRIVEN BY EXCESSIVE GOAL- is typically found that drug motivation is increased in the induction vs. no-
DIRECTED CHOICE UNDER NEGATIVE AFFECT? induction condition (sometimes also relative to pre-induction), demonstrating
the ability of negative states to motivate drug seeking [7, 10, 301, 302].
Studies with laboratory animals
The outcome-devaluation task in laboratory animals: Animals first learn an
This section will consider animal studies that test whether instrumental response to produce an appetitive reward. The response could be
dependence vulnerability is due to greater expected drug value goal-directed (driven by an expectation of current reward value), or habitual
driving goal-directed drug seeking, particularly in negative states (driven directly by an S−R association). To test this, the reward is devalued (in
one group) by pairing it with lithium chloride-induced sickness or specific satiety
[43]. Drug value can be measured by giving laboratory animals a
(in a separate context so that the test context is not paired with the devaluation
mutually exclusive choice between a response that earns the treatment). Then, at test, animals again have the opportunity to perform the
drug and a response that earns another reward such as food. instrumental response in extinction (so experience of the reinforcer cannot
The proportion of drug choices quantifies the relative value of modify any S−R association controlling the response). The response is deemed
the drug. Only some ‘vulnerable’ animals show preferential drug goal-directed in being controlled by an expectation of the current value of the
reward if the devalued group decreases responding in the extinction test. If they
choice [44–46]. Drug choice can be increased by extended drug do not show a devaluation effect, the response is deemed habitual in being
exposure [47, 48], and modified by manipulating the relative controlled directly by an S−R association. Four versions of this method have
magnitude, delay or effort associated with the two rewards been used to test whether drugs promote habitual behaviour, as outlined in the
[49–59] and by the opportunity for social interaction [60, 61]. text [30, 174–176].
The outcome-devaluation task in humans: Typically, participants learn that two
Thus, drug choice is modified by individual differences, and responses (R1 and R2) earn different rewarding outcomes (O1 and O2). One
multiple decision parameters relevant to that choice. outcome is then devalued by consumption to satiety or instructions that an
The claim that animals make goal-directed choices between outcome is no longer available. Finally, choice between the two responses is
drug and food based on the expected relative value of the tested in nominal extinction (i.e. instructions that outcomes will not be signalled
until the end). A decrease in responding for the devalued outcome suggests
rewards is supported by two lines of evidence. First, rats that that the response is goal-directed in being controlled by an expectation of the
preferentially choose drug over food have a greater number of current value of the outcome. But if responding for the devalued outcome does
neurons in the orbitofrontal cortex (OFC), which selectively not decrease (relative to baseline, or non-devalued control group), then the
activate prior to performance of the drug choice, as if the ramping response is deemed habitual in being elicited directly by an S−R association
[111, 130, 199–203, 303].
up of OFC neuronal activity is the genesis of the drug choice

Neuropsychopharmacology (2020) 0:1 – 16

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
3
The two-stage task in humans: In each trial, selecting one stimulus from the first- raising the expected value of the drug, or through a more
stage pair produces a ‘common’ and ‘rare’ second-stage pair with a 70:30 automatic form of control [96]. Furthermore, a wide range of non-
probability, respectively. If the other stimulus from the first-stage pair is selected, stress variables prompt reinstatement suggesting it may be a
the probabilities of the second-stage pairs are reversed. Selecting a second-stage
stimulus pays an amount that varies slowly over trials independently for each noisy index of emotional control of behaviour [97]. Finally, one
stimulus. Payoff is maximised by learning the transitional structure between study has found that yohimbine-induced stress did not increase
stages and which second-stage stimulus currently pays most. The goal-directed drug over food choice (Ahmed et al., personal communication
(model-based) vs. habitual (model-free) status of responding is determined by 2019). In sum, the studies reviewed here provide suggestive but
the choice of first-stage stimulus following a trial where a stimulus from the rare
second-stage pair paid most. Goal-directed participants will choose the other preliminary evidence that drug seeking in laboratory animals can
first-stage stimulus than they chose in the previous trial, giving a 70% chance of be goal-directed in some conditions, that withdrawal (and
producing the same second-stage pair as the previous trial, to access the second- possibly stress) may motivate goal-directed drug seeking, and
stage stimulus that paid most. By contrast, habitual participants will choose the that individual variation in this motivational effect could underpin
same first-stage stimulus as they chose on the previous trial because that
previous trial was reinforced, even though this choice gives only a 30% chance of dependence vulnerability. However, more work is needed to
producing the same second-stage pair as the previous trial. In short, the task convincingly determine that addiction in laboratory animals is
measures whether participants make choices using knowledge of the transitional driven by excessive goal-directed drug choice under negative
structure between stages, or simply repeat choices that paid off in the previous affect.
trial [229, 233, 278–281].
Shock punishment of self-administration in laboratory animals: Animals are first
trained on a drug self-administration schedule (or seeking-taking chain, where an Human studies
initial response is required to access the self-administration lever). After training, This section will consider human evidence that drug dependence
the self-administration response is punished by foot shock to quantify the is associated with excessive goal-directed drug choice, especially
decrease in responding relative to baseline and/or a no-punishment group.
Compulsivity is indexed by less shock suppression of self-administration
under negative affect. Dependence symptom severity is consis-
[60, 197, 283–295]. tently associated with greater economic demand (willingness to
pay) for drugs, in both non-clinical [98–100] and clinical samples
[101–103]. Economic drug demand also predicts treatment
outcomes [99, 104], and drug consumption [105], and is increased
[62, 63]. The OFC also carries signals reflecting multiple dimen- by withdrawal [106, 107], stress induction [108], impulsivity [109],
sions of rewards such as magnitude, effort, delay etc., suggesting depression, anxiety [110] and schizophrenia [102]. To the extent
the OFC may be important in calculating the overall utility of that drug demand reflects expected drug value, these studies
rewards [64]. Finally, although there is a question of homology support the claim that goal-directed drug seeking increases with
[65], the OFC plays a role in goal-directed decision making in dependence and negative affect states (withdrawal, stress
humans [66], which may translate to animals. induction, psychiatric symptoms).
The second line of evidence comes from the outcome- Excessive drug value indexed in human concurrent choice tasks
devaluation procedure used to determine whether behaviour is is also associated with dependence. Participants make forced
goal-directed [67] (see Box 1). In one study [68], rats were trained choices between drug and natural reinforcer over a series of trials.
on a seeking-taking chain for cocaine before the taking lever was Different designs use points for rewards [111, 112], pictures of
extinguished. This manipulation immediately reduced perfor- rewards [7, 113–117] or consumption of rewards [14, 118–124].
mance of the seeking response tested in extinction (in the Preferential drug choice is reliably associated with the severity of
absence of the taking lever), suggesting the seeking response was dependence to heroin [125], cocaine [116, 117, 126, 127], alcohol
controlled by a goal-directed expectation of access to cocaine, [10, 15, 26, 112, 113, 115], and tobacco [15, 111, 114, 124, 128].
rather than an S−R association. So although drug seeking can be These associations have been found in both clinical
goal-directed (see also [69–71]), as noted in the habit section, [15, 113, 114, 116, 117, 125–127] and non-clinical samples
most animal studies suggest it is habitual (see also [72]). It remains [10, 26, 111–113, 115, 128]. Percent drug choice also increases
unclear what the optimal parameters are for detecting goal- with latency to relapse [129], abstinence [7], depression and
directed vs. habitual drug-seeking behaviour [73]. anxiety symptoms and self-reported drinking to cope with
The most important question is whether negative states such as negative affect [10, 15, 113, 115], and is decreased by health
withdrawal and stress can motivate goal-directed drug seeking. warnings and satiety [111, 130, 131], by raising the magnitude of
The key study testing this prediction [74] found that when shifted the alternative reward [14, 112, 118, 121, 132–134], and by
to a state of heroin withdrawal, rats immediately increased their increasing the effort [59], and delay of the drug choice
heroin seeking in extinction, suggesting withdrawal raised the [112, 133, 134]. Thus, like economic demand, concurrent choice
expected value of heroin as a goal. Relatedly, other animal studies tasks index drug value, and this is increased in individuals with
have demonstrated that withdrawal or conditioned withdrawal dependence and associated psychiatric risk factors.
motivate drug vs. food choice, or reinstate drug self-administra- The crucial question is whether drug choice is goal-directed as
tion, or activate negative emotional brain circuits. However, these opposed to automatic. In support of the goal-directed account,
motivational effects may not necessarily be goal-directed, there drug choice is immediately reduced in extinction by decreasing
are several null effects to consider, and it is possible that the the value of the drug via satiety [11, 111, 135], pharmacotherapy
motivational impact of withdrawal may differ between drug [130] and health warnings [111, 135], indicating that drug choice is
classes [75–82]. goal-directed (see outcome-devaluation task in Box 1). The
The stress-induced reinstatement model has also produced implication is that preferential drug choice in dependent
mixed support for the goal-directed account (see Box 1). individuals is controlled by greater expected drug value.
Behavioural and pharmacological stress induction procedures Crucially for the theoretical model, goal-directed drug choice
reliably increase single lever drug seeking and taking in the has also been augmented by negative affect induction in two
reinstatement model [83, 84]. Furthermore, sensitivity to this effect studies. In the first study, smokers were trained on a concurrent
is increased by various ‘vulnerability’ factors: specifically, long vs. choice task to earn tobacco and food points, before tobacco was
short access to drugs [85–88]; adolescent vs. adult onset of drug devalued by specific satiety [11]. Participants then completed
exposure [89, 90]; oestrous cycle [91]; individual heterogeneity either a negative or positive mood induction procedure before
[92]; protracted withdrawal [93]; and baseline anxiety level at test choice was measured again in extinction. For participants in the
[94]. This work suggests that sensitivity to negative-affect-driven positive mood induction group, satiety decreased goal-directed
drug seeking may play a role in vulnerability to dependence [95]. tobacco choice as expected [111, 135]. By contrast, participants in
However, it is not clear whether stress augments drug seeking by the negative mood induction group who reported an increase in

Neuropsychopharmacology (2020) 0:1 – 16

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
4
negative mood actually increased their goal-directed tobacco sessions. Then, in separate tests, each outcome is devalued and
seeking, despite smoking satiety. The implication is that negative response rate for the devalued outcome is measured in extinction.
mood is a powerful motivational state driving goal-directed Four such designs have found that the drug-seeking response
tobacco seeking that can outweigh the primary motivational state does not decrease from baseline following devaluation, suggest-
of satiety. In the second study, alcohol drinkers were trained on a ing the behaviour is not goal-directed (controlled by expected
concurrent choice task for alcohol and food points before being outcome value), but is an S−R habit elicited by drug paired
tested in extinction with a negative or positive affect statement contextual stimuli [30, 174–176]. The food-seeking response, by
read at random prior to each choice [10]. Negative affect contrast, is reduced by devaluation, indicating that it is goal-
statements primed an increase in goal-directed alcohol choice, directed. These four studies provide the core empirical basis for
relative to positive statements and baseline, in participants who the claim that drug seeking as opposed to natural reward seeking
reported drinking to cope with negative affect. The implication is (in laboratory animals) is especially prone to habitual control.
that negative affect augments goal-directed drug choice, and this In the second design, animals are chronically exposed to a drug
effect is magnified in those who are constitutionally predisposed (experimenter administered or consumed in the home cage), and
to use drugs to cope with negative affect. then trained on a single lever for food. Food is then devalued, and
Other studies have demonstrated that, in vulnerable individuals, the food-seeking response is tested in extinction. Eight such studies
mood/stress induction has a greater priming effect on drug have shown that, at test, food seeking is insensitive to devaluation
motivation measured by pictorial drug choice, craving, economic (habitual) in drug-exposed animals and sensitive to devaluation
demand and consumption. The stress-induced increase in drug (goal-directed) in non-drug-exposed animals [69, 177–183], although
craving predicts risk of relapse in alcohol [136–139] and cocaine- three studies reported null group differences [184–186]. The
dependent individuals [140, 141], suggesting this sensitivity is a implication is that chronic drug exposure renders reward seeking
core mechanism in addiction. Mood/stress-induced drug motiva- prone to habitual control, producing general behavioural autonomy.
tion is also amplified in individuals who self-report using drugs to However, it is not clear how habitual natural reward seeking would
cope with negative affect [10, 115, 125, 136, 137, 142–149], in lead to drug dependence.
smokers with depression symptoms [7, 150], young adult drinkers In the third design, animals are trained on a single lever for
with depression symptoms [10] and alcohol-dependent men with drug, and sensitivity to devaluation is tested after minimal vs.
anxiety symptoms [137]. Individual sensitivity to mood-induced extended training. Three studies have demonstrated that the
drug seeking also correlates with withdrawal-induced drug drug-seeking response is initially goal-directed but then becomes
seeking suggesting a common mechanism [7]. Importantly habitual with extended training [69–71]. However, because food
however, although alcohol dependence has been associated with seeking also transitions from goal-directed to habitual control with
greater mood/stress-induced drug motivation in some studies training in animals [187] (not replicated in humans [188]), these
[147, 151–153], a sizable number of other studies have reported findings do not inform us about the unique habit forming
null associations [10, 113, 115, 137, 143, 146, 148, 149, 154]. The potential of drug seeking.
implication is that although dependence severity is associated In the fourth design, animals are trained on a single lever for
with preferential goal-directed drug choice, negative affect- drug and tested for sensitivity to devaluation following a fixed
induced priming of drug motivation is predominantly linked to amount of training. These studies have revealed drug seeking to
psychiatric symptoms and drug use coping motives, and this be both goal-directed [68, 74], and habitual [189], so do not inform
second process may represent the unique additional risk factor us about the unique habit forming potential of drug seeking.
that drives addiction in those with psychiatric comorbidities and There are two main criticisms of the animal outcome-devaluation
subclinical psychiatric symptoms. model. First, habitual instrumental behaviour is generally only found
Psychiatric symptoms, abuse/trauma history and associated when animals have access to a single lever in each session
drug use to cope with negative affect are major prospective risk ([190, 191] but for one exception see [30]). By contrast, it is
factors for the development and persistence of drug dependence commonly found that when rats have concurrent access to two
[155–173]. Furthermore, as shown in Table 1, self-reported drug levers for different rewards in each session, drug seeking remains
use to cope with negative affect mediates the relationship goal-directed [73], food seeking remains goal-directed despite
between psychiatric/abuse/trauma severity and dependence chronic drug exposure [184, 192], and food seeking remains goal-
severity, in a wide range of clinical and subclinical groups. directed despite extended training [193–196]. If one accepts that
Although the majority of studies listed in Table 1 are cross- human drug users’ natural environment offers a multitude of
sectional, precluding causal inferences, they nevertheless strongly responses for different rewards, then it must be concluded that
support the hypothesis that drug dependence in a wide range of habitual behaviour seen in the animal model has minimal ecological
vulnerable groups is driven by excessive goal-directed drug choice validity and likely does not play a major role in human addictive
under negative affect. behaviour [197, 198].
The second criticism is that habitual control is fragile because
sensitivity to devaluation is immediately restored in reacquisition
IS ADDICTION DRIVEN BY HABIT LEARNING? tests where drug seeking produces the devalued reinforcer. This
According to habit theory of addiction [30, 32], repeated experience restoration of sensitivity to devaluation in reacquisition tests is found
of drug reward progressively strengthens the stimulus−response for both drug seeking [30, 174–176] and food seeking in chronically
(S−R) association between drug stimuli and the drug seeking drug-exposed animals [180, 181]. If one accepts that in human drug
responses, such that drug stimuli become able to elicit drug seeking users’ natural environment, extinction conditions rarely occur, but
directly, without an expectation of the drug and its current value. conditions comparable to reacquisition prevail (i.e. drug seeking is
Thus, drug seeking becomes less susceptible to voluntary control typically reinforced), then it must be concluded that the habitual
and decision making [30, 32]. The outcome-devaluation and two- behaviour seen exclusively in the extinction test of the animal model
stage procedures are the key sources of evidence for the habit has limited ecological validity and likely does not play a major role in
account, and these studies are reviewed now. human addictive behaviour.

Studies with laboratory animals Human studies

Outcome-devaluation designs testing habit theory of addiction in Table 2 summarises outcome-devaluation studies conducted with
animals fall into four categories. In the most compelling designs, human drug users to test habit theory. There have been 11 tests
animals learn that two responses earn drug and food in separate published in 7 papers [111, 130, 199–203]. A scan of the ‘Support

Neuropsychopharmacology (2020) 0:1 – 16

 Table 1. Studies showing that self-reported drug use to cope with negative affect mediates the relationship between psychiatric/abuse/trauma symptoms and drug dependence severity.

Paper Participants Gender Index of psychiatric/abuse/trauma symptom severity Index of drug use to cope with Index of drug/alcohol dependence severity
negative affect

[249] South African school-aged Both Childhood abuse in the Childhood Trauma Q Adolescent Coping Orientation for Problem Alcohol Use Disorder Identification Test
adolescents Experiences
[250] Canadian Aboriginal school-aged Both Hopelessness subscale of the Substance Use Risk Profile Scale Drinking Motives Q Bespoke excessive drinking questionnaire
adolescents
[251] Scottish school-aged adolescents Both Neighbourhood deprivation subscale of Scottish Index of Drinking Motives Q Bespoke measure of drinking frequency
Multiple Deprivation
[252]. Armed service member reservists Both Psychological distress in the Kessler K-6 Q Drinking Motives Q Alcohol Use Disorders Identification Test
and veterans
[253]. Veterans Both PTSD measured by DSM-V Drinking Motives Q Brief Young Adult Alcohol Consequences Q
[254] University students Both Frequency of intimate partner violence Drinking Motives Q Rutgers Alcohol Problems Index

Neuropsychopharmacology (2020) 0:1 – 16

[255] University students Both Childhood abuse in the Childhood Trauma Q Marijuana Motives Measure Marijuana Problems Scale
[256] University students Both Neurotic personality: hopelessness and anxiety sensitivity Drinking Motives Q Alcohol Use Disorder Identification Test
subscales of the Substance Use Risk Profile Scale
[257] University students Both Avoidance subscale of the Liebowitz Social Anxiety Scale Drinking Motives Q Rutgers Alcohol Problem Index
[258] Adult moderate drinkers Both Negative mood in Positive and Negative Affect Scales Drinking Motives Q Bespoke measure of acute alcohol use disorder
symptoms
[259] Prisoners with history of childhood Female Trauma symptoms checklist Drinking Motives Q (modified for drugs) Inventory of Drug Use Consequences
sexual abuse
[260] Adult community mental health Both PTSD Symptom Scale Interview version Drinking Context Scale First two items of Alcohol Use Disorders
clients Identification Test
[261] School-aged adolescents Both Bespoke bullying Q Drinking Motives Q Rutgers Alcohol Problem Index
[262] University students Both Frequency of intimate partner sexual coercion Drinking Motives Q Rutgers Alcohol Problems Index
[263] University students Both Beck Depression Inventory Drinking Motives Q Brief Young Adult Alcohol Consequences Q
[264] University students Both Childhood abuse in the Childhood Trauma Q Drinking Motives Q Alcohol Use Disorder Identification Test
[265] Homeless Female PTSD symptoms in Posttraumatic Diagnostic Scale Drinking Motives Q Timeline follow back of alcohol use frequency
[266] University students with history of Both The Adult Suicidal Ideation Q Drinking Context Scale Alcohol consumption in NIAAA Questions + Young
L Hogarth

suicidal ideation Adult Alcohol Consequences Q

[267] Sample from Virginia Adult Both Lifetime major depression in DSM-IV Alcohol Use Inventory Lifetime Alcohol dependence in DSM-IV
Twin study
[268] Prisoners Female Perceived Stress Scale Substance Abuse Treatment Drug Abuse Screening Test
Functionality Scale
[269] Woman exposed to domestic Female Trauma Symptom Inventory Drinking Motives Q Timeline follow back: Number of heavy drinking days
violent abuse
[270] Adult community members Both Depression symptoms in DSM-IV Drinking Motives Q Alcohol dependence symptoms in DSM-IV
[271] Woman who have experienced Female Bespoke lifetime trauma exposure Q Two alcohol items from the Brief COPE Q Bespoke problem drinking Q
adult sexual assault
[272] Adult community members Female Single item endorsing childhood sexual assault vs. not Drinking Motives Q Bespoke alcohol dependence Q
[273] Prisoners Male Antisocial behaviour in Psychopathy Checklist—Revised Drinking Motives Q Michigan Alcoholism Screening Test —Short
[167] Adult community members Both Depression/anxiety symptoms Drinking Motives Q Bespoke drinking frequency and problems Q
[274] University students Both Anxiety Sensitivity Index Drinking Motives Q Bespoke measure of drinking frequency
[275] Adult community members Female Court-substantiated cases of child abuse and neglect vs. Single item assessing use of alcohol to cope Diagnosis of alcohol dependence
matched controls
[276] Adult community members Both Center for Epidemiologic Studies Depression Scale Drinking Motives Q Alcohol consumption frequency in the National Health
and Leisure Time Survey
[277] Adult regular gamblers Both Shame subscale of the State Shame and Guilt Scale Coping subscale of the Gambling Motives Problem Gambling Severity Index
Questionnaire

This mediational role is observed in clinical and non-clinical samples, both genders, across age groups, countries, drug classes (including gambling), and with a wide range of questionnaires (Q). These studies
suggest that excessive goal-directed drug seeking under negative affect (indexed by coping motive) drives addiction in vulnerable groups
Addiction is driven by excessive goal-directed drug choice under negative. . .

5
 6
Table 2. Outcome-devaluation procedure testing habit theory of addiction.

1 2 3 4 5 6 7 8
Test # Paper Participants Task design Reduced devaluation Reduced devaluation with Status of explicit contingency Support for
in users vs. controls? dependence severity in knowledge in drug user group habit theory
user group?

1 [111] expt. 1 Young adult Concurrent choice of tobacco and chocolate — No Accurate No
smokers points. Devaluation by specific satiety.
2 [111] expt. 2 Young adult Concurrent choice of tobacco and chocolate — No Accurate No
smokers points. Devaluation by health warnings.
3 [130] Young adult Concurrent choice of tobacco and chocolate — No Accurate No
smokers points. Tobacco devalued by 1 mg intranasal
nicotine replacement therapy.
4 [199] Young adult Concurrent choice of water vs. chocolate — No Accurate No
smokers points. Devaluation by specific satiety.
5 [200] expt. 1 Treatment-seeking Concurrent choice of water vs. crisp points. No — Accurate No
users vs. controls Devaluation by specific satiety.
6 [200] expt. 2 Treatment-seeking Concurrent choice of coca-cola vs. chocolate No — Accurate No
users vs. controls points. Devaluation by specific satiety.
7 [203] task 2 Cocaine- Concurrent choice of correct vs. incorrect No — Accurate No
dependent vs. avoidance response based on stimulus
controls signalling a shock to left or right wrist.
Devaluation by disconnecting one wrist.
L Hogarth
Addiction is driven by excessive goal-directed drug choice under negative. . .

8 [201] task 2 Heavy smokers vs. Comparable task to [203] task 2 above. No No Accurate No
controls
9 [202] Alcohol- Fabulous fruit task. Yes Impaired S-R knowledge in users vs. Yes
dependent vs. controlsa
controls
10 [203] task 1 Cocaine- Modified fabulous fruit task. Yes — Impaired S-O, R-O and S-R Yes
dependent vs. knowledge in users vs. controls
controls
11 [201] task 1 Heavy smokers vs. Modified fabulous fruit task. No Yes Not impaired in users vs. controls, Yes
controls but impaired S-R and S-O
knowledge with dependence
severity in users
Habit theory predicts reduced impact of devaluation on choice in drug users vs. controls, and/or as a function of dependence severity in the drug user groups, suggesting propensity to habit and/or impaired
goal-directed control. Columns 5, 6 and 8 highlight that only 3 out of 11 tests supported the predictions of habit theory. Furthermore, column 7 shows a strong correspondence between impaired devaluation
performance and impaired explicit contingency knowledge. In the three tests where devaluation performance was impaired (tests 9–11), explicit task contingency knowledge was also impaired, whereas in the
eight tests where devaluation performance was intact (tests 1–8) explicit contingency knowledge was also intact, suggesting that the apparent evidence for habit may be due to impaired knowledge of task
contingencies stemming from general task disengagement
a
Note that in this paper, contingency knowledge data were not published originally, but are reported in the supplementary materials of the current paper

Neuropsychopharmacology (2020) 0:1 – 16

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
7
for habit theory’ column indicates that these tests yield an 8 to 3 multiple regression analysis suggested that weaker knowledge of
ratio of evidence against habit theory. Additionally, the analysis the stimulus–outcome contingencies explained the relationship
below indicates that the three positive tests can be explained by between nicotine dependence severity and weaker devaluation
general task disengagement producing impaired explicit knowl- performance, leading the authors to conclude that “habitual
edge of task contingencies, rather than a specific propensity to responding in severely dependent smokers may be the result of
habit learning. compromised goal-directed learning”. However, because this
All the studies in Table 2 used an outcome-devaluation task in specific impairment in stimulus−outcome knowledge was not
which there was a concurrent choice between two responses that found consistently across the three studies supporting habit
earned different rewards. These rewards could be points for tobacco, theory, this conclusion cannot be maintained. Indeed, test 9 [202]
food, soft drinks, or money plus a specific outcome picture, shock or did not publish the explicit contingency knowledge data, but later
aversive noise. Outcomes were then devalued by specific satiety, by analysis of these data (reported in the supplementary materials of
instructing participants that outcome pictures would not earn the current paper) found less accurate explicit knowledge of
money points, or by disconnecting the shock or noise. Finally, stimulus−response contingencies in alcohol dependent vs.
participants were tested for choice between responses in extinction. control participants—the very knowledge that should be impor-
If participants show reduced choice of the response that led to the tant for habit learning. In sum, all three studies showing impaired
now devalued outcome, they are goal-directed, but if they do not devaluation performance also showed impaired explicit contin-
reduce choice of the devalued outcome, they are habitual. gency knowledge suggesting that the apparent evidence for habit
Habit theory predicts that the effect of devaluation on choice at could be explained by general task disengagement perhaps
test should be reduced in drug users vs. controls, and/or as a driven by a general cognitive impairment [37, 210–212].
function of dependence severity in the user group, demonstrating The idea that devaluation performance could be disrupted by
a propensity to habit. Columns 5 and 6 of Table 2 summarise general cognitive impairment is supported by ‘cognitive load’
evidence for these predictions. Tests numbered 1–8 provide no studies in humans and animals. These studies have found that
evidence for habit theory, in that the devaluation effect was not devaluation performance can be impaired by stress [214–219],
reduced in the drug user group or as a function of dependence acute alcohol administration [220], an alcohol consumption
severity. This failure to support habit theory was found in both expectancy [135], being placed in drug-related contexts
clinical (tests 5–8) and non-clinical samples (1–4). Furthermore, the [221, 222], and sleep deprivation [223]. Furthermore, devaluation
failure to support habit theory cannot be attributed to the use of performance is impaired in a range of neuropsychiatric conditions
concurrent choice procedures (which tend to discourage habitual including social anxiety [224, 225], autism spectrum disorder [225],
learning as noted earlier) because all of the tests in Table 2 used schizophrenia [226], Parkinson’s disease [227], obsessive compul-
concurrent choice procedures, both those that failed (test 1–8) sive disorder [213], impulsivity [199], and in young children [228].
and those that notionally supported habit theory (test 9–11). The generality of the devaluation deficit suggests it stems from
In those tests that supported habit theory (tests 9–11), the drug general motivational or cognitive impairments, and is not the
user group or more dependent users also showed impaired unique mechanism underpinning addiction.
explicit knowledge of the contingencies operating in the task, in A similar analysis may be applied to the two-stage task (see
addition to weaker devaluation performance. Indeed, column Box 1 for a description of the methods). The results of these
7 shows a perfect correspondence between impaired explicit studies are summarised in Table 3. There have been nine tests of
contingency knowledge and impaired devaluation performance. whether model-based (goal-directed) learning is impaired or
In numerous human learning tasks, explicit knowledge of task model-free (habit) learning is increased in drug users vs. controls
contingencies is necessary for accurate performance, that is, (column 4), or as a function of dependence severity in the user
participants who have impaired contingency knowledge perform group (column 5). Of the nine tests, only four claimed evidence for
less accurately in these tasks [204–209]. The implication is that habit theory (although one of these was one-tailed and the group
drug users, or more dependent users, failed to acquire explicit difference was not significant when a confound in cognitive speed
contingency knowledge in the three tests supporting habit theory was controlled [229]). Crucially, none of the studies measured
(tests 9–11), which impaired their devaluation performance, participants’ explicit knowledge of the task contingencies.
making them appear to be habitual. Drug users have general Consequently, it is unknown whether the four studies reporting
deficits in cognition or motivation that underpin their perfor- evidence for habit theory can be explained by impaired
mance deficits in wide range of tasks [37, 210–212]. Arguably, this contingency knowledge (as was the case with the outcome-
general deficit in cognition or motivation produced general task devaluation procedure). However, it is known that model-based
disengagement that impaired explicit contingency knowledge learning can be increased by adding incentives (points) for
and thereby impaired devaluation performance. In other words, accurate performance [230], and impaired by a working memory
impaired devaluation performance is probably not driven by a load manipulation [231], and is impaired in individuals with lower
specific propensity to habit learning or deficit in goal-directed working memory capacity [232], and lower cognitive speed
control but by general task disengagement. [233, 234]. These findings suggest that the four two-stage studies
It is important to note that drug users’ deficit in explicit that reported evidence for habit theory may be attributed to
contingency knowledge extended to knowledge of stimulus general motivational/cognitive deficit in drug users or as a
−response (S−R), response−outcomes (R−O) and stimulus−out- function of dependence severity, rather than a specific propensity
come (S−O) contingencies. Thus, it cannot be claimed that drug to habit learning or deficit in goal-directed control. To quote one
users were specifically impaired in learning about outcomes two-stage paper [233]: “whether reduced model-based control in
important for goal-directed action, as has sometimes been patients constitutes a disease-specific mechanism or results from
claimed [201, 213]. For example, the supplemental material for general cognitive impairments can only be teased apart in future
test 10 [203] states that: “Compared with control volunteers, CUD longitudinal studies”.
[cocaine dependent participants] demonstrated significant deficits
in explicit knowledge in terms of stimulus-outcome … response-
outcome … and stimulus-response … relationships”. Similarly, in IS ADDICTION DRIVEN BY COMPULSION (INSENSITIVITY TO
test 11 [201], “Strong evidence was obtained for a negative PUNISHMENT)?
association between FTND scores [nicotine dependence severity Studies with laboratory animals
in smokers] and explicit knowledge on stimulus–response… One of the major problems with habit theory noted earlier is that
and stimulus–outcome associations”. Furthermore, hierarchical sensitivity to devaluation is restored immediately when responses

Neuropsychopharmacology (2020) 0:1 – 16

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
8
produce the devalued reinforcer, so habit could not explain the

Habit theory predicts reduced model-based or increased model-free learning in drug users vs. controls, or with dependence severity in the user groups, suggesting impaired goal-directed control/propensity to
habit. Columns 4 and 5 highlight that only four out of nine tests have supported these predictions. Furthermore, one of the positive studies (test 6 [229]) was rendered null when the group difference in
cognitive capacity was controlled. It remains unclear whether the other positive results stem from general motivational or cognitive deficits, or tap the prospective mechanism underpinning dependence
persistence of drug seeking in the human natural environment. To
negate this theoretical dilemma, it has been proposed that drug
habit theory seeking is controlled by compulsion, defined as “a maladaptive
Support for
stimulus-response habit” [31], and “as the maladaptive persistence
of responding despite adverse consequences” [32] (it should

Yes/no
be noted that other researchers use the term compulsion to mean

Yes

Yes

Yes
No
No

No

No
No
a wide range of processes which are not considered here because
6

they are difficult to test empirically [198]). The principal assay of

model-free learning in users vs. controls learning with dependence severity in user group

compulsivity is the persistence of punished drug seeking. The

problem, however, is that persistence of punished drug seeking
Reduced model-based/increased model-free

can equally be explained by excessive value ascribed to the drug,

outweighing the punisher [43]. Unique evidence for the compul-
sion model relies on demonstrating that persistence of punished
drug seeking is not associated with excessive valuation of the
drug in another assay. Early studies presented preliminary support
for this dissociation between assays, but later studies that have
employed more sensitive measures of drug value have indicated
that persistence of punished drug seeking is associated with
greater valuation of the drug, undermining the core behavioural
evidence for compulsion theory.
Table 4 summarises studies that have measured the suppres-
sion of drug self-administration by shock punishment (the
putative assay of compulsivity), and drug value in a second assay
(e.g., self-administration frequency, breakpoints in progressive
Yes

Yes
No
No

No
No

No
—

ratio tasks, persistence under extinction, and preferential choice of

5

drug vs. natural reward). Studies reporting a dissociation between

these two assays (column 5) support compulsion theory by
suggesting that persistence under punishment cannot be
Reduced model-based/increased

explained by greater drug value. By contrast, studies reporting a

correlation between these two assays contradict compulsion
theory by suggesting that persistence under punishment may be
due to greater drug value. Column 5 indicates that four studies
support compulsion theory, and 11 studies support drug value
theory. Overall, the evidence favours drug value over compulsion
theory as an explanation for the persistence of punished drug
seeking. One speculation is that vulnerable animals persist under
punishment not because they are insensitive to costs, but because
the punisher motivates drug self-administration in the following
Yes
Yes
No
No

No

—
—

General online sample, with alcohol dependence —

period to self-medicate. If this is true, then persistence of punished

4

drug seeking could be another example of excessive goal-directed

University students with substance use as a key

University students with substance use as a key

drug seeking under negative affect, not an example of

compulsion.
[278] comparison 1 Methamphetamine dependent vs. controls
[278] comparison 2 Currently abstinent alcohol-dependent vs.

Children of alcoholic fathers vs. controls

Human studies
Two-stage procedures testing habit theory of addiction.

Human studies designed to test whether dependence is

associated with insensitivity to costs (cost discounting) have also
Alcohol-dependent vs. controls
Alcohol-dependent vs. control

provided minimal evidence for compulsion theory. Demand tasks

measure the amount of drug participants would hypothetically
18-year-old social drinkers

severity as a key measure

consume across increasing prices (costs). The intensity of demand

(maximum consumption at low price) is considered to be a
relatively pure index of drug value unaffected by costs. By
contrast, breakpoint—the price at which drug consumption drops
to zero—is thought to be more sensitive to the impact of price
Participants

costs on the decision to consume. Compulsion theory would be

measure

measure
controls

supported if dependence severity was more strongly associated

with breakpoint than intensity, suggesting cost insensitivity is
more important than drug value [235–237]. However, meta-
3

analyses and systematic reviews of this literature have found that

proxies for dependence correlate more consistently with measures
of intensity than breakpoint [238–241], suggesting that depen-
[281] expt. 2

[281] Expt. 1

dence is more likely to be driven by greater drug value than cost

discounting. However, one key study found that student drinkers
Test # Paper

with a family history of alcoholism were less sensitive to the effect

[233]
[279]
[280]

[229]

[282]

of imagined next-day responsibilities on reducing alcohol demand

2
Table 3.

[242], supporting the notion that dependence vulnerability may

be linked to discounting costs imposed on alcohol. It remains to
1

2
3
4
5

6
7
8

be seen what explains this discrepancy.

Neuropsychopharmacology (2020) 0:1 – 16

 Table 4. Studies that have tested compulsion theory of addiction by determining whether persistence of punished drug seeking is dissociated from drug value measured in another assay.

1 2 3 4 5
Test # Paper Criteria to identify vulnerable vs. nonvulnerable groups Measure of drug value Relation between assay of punishment suppression and
drug value?

1 [283] High impulsivity Acquisition of single lever self-admin dose−response function Dissociated
2 [284] Long access Acquisition and reacquisition of seeking-taking chain for Dissociated
single dose
3 [285] N/A Higher breakpoints for self-admin. Dissociated
4 [286] N/A Higher breakpoints for self-admin. Dissociated
5 [287] Long access Progressive ratio breakpoint Correlated
6 [288] High novelty preference Progressive ratio breakpoint and persistence under extinction Correlated

Neuropsychopharmacology (2020) 0:1 – 16

7 [289] N/A Economic demand measure akin to progressive ratio Correlated
8 [290] Long access Acquisition and reacquisition of seeking-taking chain for Correlated
single dose
9 [291] N/A Economic demand measure akin to progressive ratio Correlated
10 [292] 3 Crit model Progressive ratio breakpoint and persistence under extinction Correlated
11 [293] 3 Crit model Progressive ratio breakpoint & persistence under extinction Correlated
12 [197] 3 Crit model Progressive ratio breakpoint and persistence under extinction Correlated
13 [294] 3 Crit model Progressive ratio breakpoint and persistence under extinction Correlated
14 [60] 3 Crit model Progressive ratio breakpoint and persistence under extinction Correlateda
15 [295] N/A Choice of drug over natural reward Correlated
The top four studies report a dissociation between these assays suggesting persistence of punished drug seeking has a different mechanism to drug value (or the design was not sensitive enough to detect the
correlation). The bottom 11 studies reported a correlation between persistence of punished drug seeking and greater drug value, suggesting persistence may be driven by drug value outweighing costs. Overall,
the evidence favours drug value over compulsion account. Several technical details of the studies are noteworthy. The majority of studies identified separate groups of animals as vulnerable vs. nonvulnerable to
dependence using various criteria described in column 3 (e.g., impulsivity or the 3 crit model [296]). If vulnerable vs. nonvulnerable animals showed less punishment suppression but no difference in drug value,
L Hogarth

the measures were defined as dissociated. But if vulnerable vs. nonvulnerable animals showed less punishment suppression and greater drug value, the measures were defined as correlated (sometimes these
studies also reported the correlation coefficient between punishment suppression and drug value, which corroborated the conclusion from the group contrasts). Other studies included a single group of animals
(in which case column 3 was labelled as N/A), and reported the correlation coefficient between punishment suppression and drug value. Column 5 was labelled as ‘correlated’ if this relationship was positive and
significant. Finally, column 4 labels the method used to measure drug value. There are a multitude of procedural parameters that could explain differential sensitivity of the measure of drug value between
experiments
a
Note there was no correlation between persistence of punished self-administration and choice of drug over social reinforcement in this case
Addiction is driven by excessive goal-directed drug choice under negative. . .

9
 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
10
Deficits in reversal learning have been interpreted as evidence addictive behaviour in the natural environment where these
for greater cost discounting in addiction. In reversal learning tasks, conditions prevail. In humans, most studies found no evidence
participants learn that one response has a higher payoff than an for habitual behaviour in drug users vs. controls, or as a function
alternative choice, before these response−reward contingencies of dependence. The three studies supporting habit theory
are reversed. Drug users show deficits in reversal learning despite showed a correspondence between impaired devaluation
comparable acquisition of the initial contingencies [35, 243–245] performance and impaired explicit contingency knowledge.
(for similar findings with laboratory animals see [246, 247]). One Furthermore, deficits in devaluation performance have been
interpretation is that drug users are less sensitive to the found with a wide range of ‘cognitive load’ manipulations and
punishment of the incorrect choice, driving persistence of this psychiatric symptom states, suggesting a general effect pro-
choice. However the effect could also be due to impaired duced by task disengagement. Finally, only the minority of
prediction error coding, cognitive inflexibility, or general task studies using two-stage tasks supported habit theory, and these
disengagement [35]. Reversal learning deficit therefore do not could also be explained by task disengagement. Collectively the
provide compelling evidence for cost discounting in addiction. studies provide minimal evidence for a specific propensity to
A recent study directly tested whether alcohol dependence was habit or impairment in goal-directed control as a major factor
associated with discounting delay and opportunity costs imposed on controlling human addiction.
alcohol seeking [112]. Student drinkers (n = 127, who varied in The evidence for compulsion theory is also weak. In animals, the
alcohol dependence symptom severity) made concurrent forced primary index of compulsivity—persistence of punished drug
choices between alcohol and food points under conditions that seeking in vulnerable animals—is most often associated with greater
manipulated the magnitude of points and the delay to receive drug value indexed in a separate assay. The implication is that
points. Alcohol value was indexed by preferential choice of alcohol persistence of punished drug seeking is not due to insensitivity to
vs. food, whereas sensitivity to costs was indexed by the decrease in costs imposed on drug seeking (compulsivity), but due to excessive
alcohol choice when food points were of greater magnitude value of the drug. Human studies have similarly found minimal
(sensitivity to opportunity costs) and when alcohol points were evidence that dependence is associated with cost discounting in
delayed (sensitivity to delay costs). It was found that alcohol use economic demand tasks, reversal learning tasks or concurrent choice
disorder symptom severity was associated with increased alcohol tasks where costs are imposed on the drug choice.
choice indicating greater value of alcohol, but not with sensitivity to The overall conclusion from this translational analysis is that
opportunity or delay costs imposed on the alcohol choice. This paper addiction is primarily driven by excessive goal-directed drug
provided further evidence that dependence is driven by greater choice under negative affect, and much less by habit or
value ascribed to drugs, and not with greater discounting of costs compulsion. This conclusion accords with other negative reinfor-
imposed on drugs, i.e. compulsion theory was not supported. cement models of addiction [3, 9, 248], except that in the current
model, negative states enhance the expected value of the drug
driving goal-directed drug choice [10, 11, 74], rather than priming
SUMMARY AND CONCLUSION drug seeking automatically as is commonly claimed [9]. Addiction
The paper reviewed studies with laboratory animal and humans is pathological not because it is automatic, but because negative
that tested whether addiction is driven by excessive goal-directed states powerfully drive up expected drug value acutely out-
drug choice under negative affect, habit, or compulsion. There weighing other goals such as a job, abstinence, family and health,
was substantial support for the first account, and limited support resulting in a return to drug use despite wishes to the contrary
for the latter two. Animal studies supporting the excessive goal expressed at other times.
account found that drug choice was associated with dependence
vulnerability, can be modulated by multiple decision parameters,
is subserved by the OFC decision-making centre, and is goal- FUNDING AND DISCLOSURE
directed. However, there was only indirect suggestive evidence The authors declare that they do not have any conflicts of interest
that negative states such as withdrawal and stress motivate drug (financial or otherwise) related to the content of the paper. The
seeking via a goal-directed mechanism, and that this effect might research was supported by an Alcohol Change grant (RS17/03)
be amplified in vulnerable animals. This area needs more and a Medical Research Council (UK, MRC) Confidence in Global
attention, back-translating human findings. Human studies, by Mental Health pump priming award (MC_PC_MR/R019991/1) to
contrast, supported the excessive goal account by demonstrating LH. Funders had no role in the study design, collection, analysis or
that economic drug demand increases with dependence, psy- interpretation of the data, writing the manuscript, or the decision
chiatric symptoms, and stress induction. Similarly, concurrent drug to submit the paper for publication
choice is demonstrably goal-directed, is modulated by decision
parameters, and increases with dependence, psychiatric symp-
toms, and mood/stress induction, and this latter effect is amplified ADDITIONAL INFORMATION
in individuals who report psychiatric symptoms and drug use to Supplementary information accompanies this paper at (https://doi.org/10.1038/
cope with negative affect, and in those at greater risk of s41386-020-0600-8).
relapse. Finally, psychiatric symptoms, abuse/trauma history, and
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims
coping motives confer prospective risk of dependence, and
in published maps and institutional affiliations.
coping motives mediate this risk. These data provide converging
translational (and longitudinal) evidence that addiction is primarily
driven by excessive goal-directed drug choice under negative REFERENCES
affect, although evidential gaps do need to be addressed. 1. Badiani A, Belin D, Epstein D, Calu D, Shaham Y. Opiate versus psychostimulant
By contrast, the evidence for habit theory is weak, suggesting addiction: the differences do matter. Nat Rev Neurosci. 2011;12:685–700.
that this psychological process does not play a major role in 2. Wikler A. Conditioning factors in opiate addiction and relapse. J Subst Abus
human addiction. Only four animal studies showed that drug Treat. 1984;1:279–85.
seeking is especially prone to habit, and although eight studies 3. Solomon RL, Corbit JD. An opponent-process theory of motivation: II. Cigarette
showed that drug exposure renders reward seeking habitual, it is addiction. J Abnorm Psychol. 1973;81:158–71.
unclear how this would drive dependence. Habitual behaviour is 4. Siegel S. Pharmacological conditioning and drug effects. In: Goudie A, Emmett-
also abolished when animals are given a choice, and when Oglesby M, editors. Psychoactive drugs tolerance and sensitisation. Clifton, New
Jersey: Humana Press; 1989. p. 115–80.
responses are rewarded, and so could not explain human

Neuropsychopharmacology (2020) 0:1 – 16

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
11
5. Khantzian EJ. The self-medication hypothesis of substance use disorders: a 35. Ersche KD, Roiser JP, Robbins TW, Sahakian BJ. Chronic cocaine but not chronic
reconsideration and recent applications. Harv Rev Psychiatry. 1997;4:231–44. amphetamine use is associated with perseverative responding in humans.
6. Aguirre C, Madrid J, Leventhal AM. Tobacco withdrawal symptoms mediate Psychopharmacology. 2008;197:421–31.
motivation to reinstate smoking during abstinence. J Abnorm Psychol. 36. Bickel WK, Johnson MW, Koffarnus MN, MacKillop J, Murphy JG. The behavioral
2015;124:623–34. economics of substance use disorders: reinforcement pathologies and their
7. Hogarth L, Mathew AR, Hitsman B. Current major depression is associated with repair. Ann Rev Clin Psych. 2014;10:641–77.
greater sensitivity to the motivational effect of both negative mood induction 37. Rochat L, Maurage P, Heeren A, Billieux J. Let’s open the decision-making
and abstinence on tobacco-seeking behavior. Drug Alcohol Depend. umbrella: a framework for conceptualizing and assessing features of impaired
2017;176:1–6. decision making in addiction. Neuropsychol Rev. 2019;29:27–51.
8. Mathew AR, Hogarth L, Leventhal AM, Cook JW, Hitsman B. Cigarette smoking 38. Wills TA, Simons JS, Sussman S, Knight R. Emotional self-control and dysregu-
and depression comorbidity: systematic review and proposed theoretical lation: a dual-process analysis of pathways to externalizing/internalizing
model. Addiction 2017;112:401–12. symptomatology and positive well-being in younger adolescents. Drug Alcohol
9. Baker TB, Piper ME, McCarthy DE, Majeskie MR, Fiore MC. Addiction motivation Depend. 2016;163:S37–S45.
reformulated: an affective processing model of negative reinforcement. Psychol 39. Hussong AM, Jones DJ, Stein GL, Baucom DH, Boeding S. An internalizing
Rev. 2004;111:33–51. pathway to alcohol use and disorder. Psychol Addictive Behav.
10. Hogarth L, Hardy L. Depressive statements prime goal-directed alcohol-seeking 2011;25:390–404.
in individuals who report drinking to cope with negative affect. Psycho- 40. Sussman S, Leventhal A, Bluthenthal RN, Freimuth M, Forster M, Ames SL. A
pharmacology. 2018;235:269–79. framework for the specificity of addictions. Int J Env Res Public Health.
11. Hogarth L, He Z, Chase HW, Wills AJ, Troisi J II, Leventhal AM, et al. Negative 2011;8:3399–415.
mood reverses devaluation of goal-directed drug-seeking favouring an incen- 41. Corbit LH, Balleine BW. Instrumental and Pavlovian incentive processes have
tive learning account of drug dependence. Psychopharmacology. dissociable effects on components of a heterogeneous instrumental chain. J Exp
2015;232:3235–47. Psychol Anim Behav Process. 2003;29:99–106.
12. Wise RA, Koob GF. The development and maintenance of drug addiction. 42. Di Ciano P, Everitt BJ. Differential control over drug-seeking behavior by drug-
Neuropsychopharmacology. 2014;39:254–62. associated conditioned reinforcers and discriminative stimuli predictive of drug
13. Stewart J, de Wit H, Eikelboom R. Role of conditioned and unconditioned drug availability. Behav Neurosci. 2003;117:952–60.
effects in self-administration of opiates and stimulants. Psychol Rev. 43. Kearns DN, Kim JS, Tunstall BJ, Silberberg A. Essential values of cocaine and non-
1984;63:251–68. drug alternatives predict the choice between them. Addict Biol. 2017;22:
14. Higgins ST, Bickel WK, Hughes JR. Influence of an alternative reinforcer on 1501–14.
human cocaine self-administration. Life Sci. 1994;55:179–87. 44. Ahmed SH. Validation crisis in animal models of drug addiction: beyond non-
15. Hardy L, Parker S, Hartley L, Hogarth L. A concurrent pictorial drug choice task disordered drug use toward drug addiction. Neurosci Biobehav Rev.
marks multiple risk factors in treatment-engaged smokers and drinkers. Behav 2010;35:172–84.
Pharmacol. 2018;29:716–25. 45. Panlilio LV, Hogarth L, Shoaib M. Concurrent access to nicotine and sucrose in
16. Heyman GM, Mims V. What addicts can teach us about addiction: a natural rats. Psychopharmacology . 2015;232:1451–60.
history approach. Addiction and choice. Oxford: Oxford University Press; 2016. 46. Lenoir M, Augier E, Vouillac C, Ahmed SH. A choice-based screening method for
17. Carter BL, Tiffany ST. Meta-analysis of cue-reactivity in addiction research. compulsive drug users in rats. Current protocols in neuroscience. John Wiley &
Addiction. 1999;94:327–40. Sons, Inc.; 2013.
18. Tomie A. Locating reward cue at response manipulandum (CAM) induces 47. Lenoir M, Cantin L, Vanhille N, Serre F, Ahmed SH. Extended heroin access
symptoms of drug abuse. Neurosci Biobehav Rev. 1996;20:505–35. increases heroin choices over a potent nondrug alternative. Neuropsycho-
19. Bindra D. How adaptive behaviour is produced: a perceptual-motivational pharmacology. 2013;38:1209–20.
alternative to response reinforcement. Behavioural Brain Sci. 1978;1:41–91. 48. Russo M, Funk D, Loughlin A, Coen K, Lê AD. Effects of alcohol dependence on
20. Robinson TE, Berridge KC. The neural basis of drug craving: an incentive- discrete choice between alcohol and saccharin. Neuropsychopharmacology.
sensitization theory of drug addiction. Brain Res Rev. 1993;18:247–91. 2018;43:1859–66.
21. Wiers RW, Boelema SR, Nikolaou K, Gladwin TE. On the development of implicit 49. Campbell UC, Carroll ME. Reduction of drug self-administration by an alternative
and control processes in relation to substance use in adolescence. Curr non-drug reinforcer in rhesus monkeys: magnitude and temporal effects. Psy-
Addiction Rep. 2015;2:141–55. chopharmacology. 2000;147:418–25.
22. Seabrooke T, Hogarth L, Mitchell CJ. The propositional basis of cue-controlled 50. Woolverton WL, Anderson KG. Effects of delay to reinforcement on the choice
reward seeking. Q J Exp Psychol. 2016;69:2452–70. between cocaine and food in rhesus monkeys. Psychopharmacology.
23. Hogarth L, Retzler C, Munafò MR, Tran DMD, Troisi Ii Jr, Rose AK, et al. Extinction 2006;186:99–106.
of cue-evoked drug-seeking relies on degrading hierarchical instrumental 51. Carroll ME, Lac ST, Nygaard SL. A concurrently available nondrug reinforcer
expectancies. Behav Res Ther. 2014;59:61–70. prevents the acquisition or decreases the maintenance of cocaine-reinforced
24. Perkins KA. Subjective reactivity to smoking cues as a predictor of quitting behavior. Psychopharmacology. 1989;97:23–29.
success. Nicotine Tob Res. 2012;14:383–87. 52. Carroll ME, Lac ST. Autoshaping i.v. cocaine self-administration in rats: effects of
25. Perkins KA. Does smoking cue-induced craving tell us anything important about nondrug alternative reinforcers on acquisition. Psychopharmacology.
nicotine dependence? Addiction. 2009;104:1610–16. 1993;110:5–12.
26. Hardy L, Mitchell C, Seabrooke T, Hogarth L. Drug cue reactivity involves hier- 53. Ginsburg BC, Lamb RJ. Frustration stress (unexpected loss of alternative rein-
archical instrumental learning: evidence from a biconditional Pavlovian to forcement) increases opioid self-administration in a model of recovery. Drug
instrumental transfer task. Psychopharmacology. 2017;234:1977–84. Alcohol Depend. 2018;182:33–39.
27. Field M, Werthmann J, Franken I, Hofmann W, Hogarth L, Roefs A. The role of 54. LeSage MG. Toward a nonhuman model of contingency management: effects of
attentional bias in obesity and addiction. Health Psychol. 2016;35:767–80. reinforcing abstinence from nicotine self-administration in rats with an alter-
28. Piazza PV, Deroche-Gamonet V. A multistep general theory of transition to native nondrug reinforcer. Psychopharmacology. 2009;203:13–22.
addiction. Psychopharmacol (Berl). 2013;229:387–413. 55. Nader MA, Woolverton WL. Choice between cocaine and food by rhesus
29. Tiffany ST. A cognitive model of drug urges and drug-use behaviour: role of monkeys: effects of conditions of food availability. Behavioural Pharmacol.
automatic and nonautomatic processes. Psychol Rev. 1990;97:147–68. 1992;3:635–38.
30. Dickinson A, Wood N, Smith JW. Alcohol seeking by rats: action or habit? Q J Exp 56. Nader MA, Woolverton WL. Effects of increasing response requirement on
Psychol B. 2002;55:331–48. choice between cocaine and food in rhesus monkeys. Psychopharmacology.
31. Everitt BJ, Robbins TW. Neural systems of reinforcement for drug addiction: from 1992;108:295–300.
actions to habits to compulsion. Nat Neurosci. 2005;8:1481–9. 57. Nader MA, Woolverton WL. Effects of increasing the magnitude of an alternative
32. Everitt BJ, Robbins TW. Drug addiction: updating actions to habits to compul- reinforcer on drug choice in a discrete-trials choice procedure. Psycho-
sions ten years on. Annu Rev Psychol. 2016;67:23–50. pharmacology. 1991;105:169–74.
33. Verdejo-García A, Alcázar-Córcoles MA, Albein-Urios N. Neuropsychological 58. Negus SS. Rapid assessment of choice between cocaine and food in rhesus
interventions for decision-making in addiction: a systematic review. Neu- monkeys: effects of environmental manipulations and treatment with d-
ropsychol Rev. 2019;29:79–92. amphetamine and flupenthixol. Neuropsychopharmacology. 2003;28:919.
34. Volkow ND, Fowler JS, Wang GJ. The addicted human brain viewed in the light 59. Ito M, Nakamura K. Humans’ choice in a self-control choice situation: sensitivity
of imaging studies: brain circuits and treatment strategies. Neuropharmacology. to reinforcer amount, reinforcer delay, and overall reinforcement density. J Exp
2004;47:3–13. Anal Behav. 1998;69:87–102.

Neuropsychopharmacology (2020) 0:1 – 16

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
12
60. Venniro M, Zhang M, Caprioli D, Hoots JK, Golden SA, Heins C, et al. Volitional behavioral responses are augmented following long-access cocaine self-
social interaction prevents drug addiction in rat models. Nat Neurosci. administration by rats. Psychopharmacology. 2008;195:591–603.
2018;21:1520–29. 87. Ahmed SH, Walker JR, Koob GF. Persistent increase in the motivation to take
61. Venniro M, Caprioli D, Shaham Y. Novel models of drug relapse and craving after heroin in rats with a history of drug escalation. Neuropsychopharmacology.
voluntary abstinence. Neuropsychopharmacology. 2019;44:234–35. 2000;22:413–21.
62. Guillem K, Brenot V, Durand A, Ahmed SH. Neuronal representation of individual 88. Blacktop JM, Seubert C, Baker DA, Ferda N, Lee G, Graf EN, et al. Augmented
heroin choices in the orbitofrontal cortex. Addict Biol. 2017;23:880–888. cocaine seeking in response to stress or CRF delivered into the ventral teg-
63. Guillem K, Ahmed SH. Preference for cocaine is represented in the orbitofrontal mental area following long-access self-administration is mediated by CRF
cortex by an increased proportion of cocaine use-coding neurons. Cereb Cortex. receptor type 1 but not CRF receptor type 2. J Neurosci. 2011;31:11396.
2017;28:819–32. 89. Siegmund S, Vengeliene V, Singer MV, Spanagel R. Influence of age at drinking
64. Wallis JD. Neuronal mechanisms in prefrontal cortex underlying adaptive choice onset on long-term ethanol self-administration with deprivation and stress
behavior. Linking affect to action: critical contributions of the orbitofrontal phases. Alcohol Clin Exp Res. 2005;29:1139–45.
cortex. Oxford: Blackwell Publishing; 2007. p. 447–60. 90. Fullgrabe MW, Vengeliene V, Spanagel R. Influence of age at drinking onset on
65. Balleine BW, O’Doherty JP. Human and rodent homologies in action control: the alcohol deprivation effect and stress-induced drinking in female rats. Pharm
corticostriatal determinants of goal-directed and habitual action. Neu- Biochem Behav. 2007;86:320–6.
ropsychopharmacology. 2010;35:48–69. 91. Feltenstein MW, Henderson AR, See RE. Enhancement of cue-induced rein-
66. Valentin V, Dickinson A, O’Doherty JP. Determining the neural substrates of statement of cocaine-seeking in rats by yohimbine: sex differences and the role
goal-directed learning in the human brain. J Neurosci. 2007;27:4019–26. of the estrous cycle. Psychopharmacology (Berlin). 2011;216:53–62.
67. Dickinson A, Balleine BW. The cognitive/motivational interface. In: Kringelbach 92. Buffalari DM, Baldwin CK, Feltenstein MW, See RE. Corticotrophin releasing
ML, Berridge KC, editors. Pleasures of the brain. The neural basis of taste, smell factor (CRF) induced reinstatement of cocaine seeking in male and female rats.
and other rewards. Oxford: Oxford University Press; 2010. p. 74–84. Physiol Behav. 2012;105:209–14.
68. Olmstead MC, Lafond MV, Everitt BJ, Dickinson A. Cocaine seeking by rats is a 93. Shalev U, Morales M, Hope B, Yap J, Shaham Y. Time-dependent changes in
goal-directed action. Behav Neurosci. 2001;115:394–402. extinction behavior and stress-induced reinstatement of drug seeking following
69. Corbit LH, Nie H, Janak PH. Habitual alcohol seeking: time course and the withdrawal from heroin in rats. Psychopharmacology. 2001;156:98–107.
contribution of subregions of the dorsal striatum. Biol Psychiatry. 94. Erb S. Evaluation of the relationship between anxiety during withdrawal and
2012;72:389–95. stress-induced reinstatement of cocaine seeking. Prog Neuropsychopharmacol
70. Zapata A, Minney VL, Shippenberg TS. Shift from goal-directed to habitual Biol Psychiatry. 2010;34:798–807.
cocaine seeking after prolonged experience in rats. J Neurosci. 95. Heilig M, Egli M, Crabbe JC, Becker HC. Acute withdrawal, protracted abstinence
2010;30:15457–63. and negative affect in alcoholism: are they linked? Addict Biol. 2010;15:169–84.
71. Clemens KJ, Castino MR, Cornish JL, Goodchild AK, Holmes NM. Behavioral and 96. Schwabe L, Dickinson A, Wolf OT. Stress, habits and drug addiction: a psycho-
neural substrates of habit formation in rats intravenously self-administering neuroendocrinological perspective. Exp Clin Psychopharmacol. 2011;19:53–63.
nicotine. Neuropsychopharmacology. 2014;39:2584–93. 97. Girardeau P, Navailles S, Durand A, Vouillac-Mendoza C, Guillem K, Ahmed SH.
72. Vandaele Y, Vouillac-Mendoza C, Ahmed SH. Inflexible habitual decision-making Relapse to cocaine use persists following extinction of drug-primed craving.
during choice between cocaine and a nondrug alternative. Transl Psychiatry. Neuropharmacology. 2019;155:185–93.
2019;9:109. 98. Murphy JG, MacKillop J. Relative reinforcing efficacy of alcohol among college
73. Halbout B, Liu AT, Ostlund SB. A closer look at the effects of repeated cocaine student drinkers. Exp Clin Psychopharmacol. 2006;14:219–27.
exposure on adaptive decision-making under conditions that promote goal- 99. MacKillop J, Murphy JG. A behavioral economic measure of demand for alcohol
directed control. Front Psychiatry. 2016;7:44. predicts brief intervention outcomes. Drug Alcohol Depend. 2007;89:227–33.
74. Hutcheson DM, Everitt BJ, Robbins TW, Dickinson A. The role of withdrawal in 100. Murphy JG, MacKillop J, Skidmore JR, Pederson AA. Reliability and validity of a
heroin addiction: enhances reward or promotes avoidance? Nat Neurosci. demand curve measure of alcohol reinforcement. Exp Clin Psychopharmacol.
2001;4:943–47. 2009;17:396–404.
75. Kenny PJ, Chen SA, Kitamura O, Markou A, Koob GF. Conditioned withdrawal 101. Petry NM. A behavioral economic analysis of polydrug abuse in alcoholics:
drives heroin consumption and decreases reward sensitivity. J Neurosci. asymmetrical substitution of alcohol and cocaine. Drug Alcohol Depend.
2006;26:5894–900. 2001;62:31–9.
76. Shaham Y, Rajabi H, Stewart J. Relapse to heroin-seeking in rats under opioid 102. MacKillop J, Tidey JW. Cigarette demand and delayed reward discounting in
maintenance: the effects of stress, heroin priming, and withdrawal. J Neurosci. nicotine dependent individuals with schizophrenia and controls: an initial study.
1996;16:1957. Psychopharmacology. 2011;216:91–99.
77. Negus SS, Banks ML. Modulation of drug choice by extended drug access and 103. Bruner NR, Johnson MW. Demand curves for hypothetical cocaine in cocaine-
withdrawal in rhesus monkeys: implications for negative reinforcement as a dependent individuals. Psychopharmacology. 2014;231:889–97.
driver of addiction and target for medications development. Pharm Biochem 104. Murphy JG, Correia CJ, Colby SM, Vuchinich RE. Using behavioral theories of
Behav. 2018;164:32–39. choice to predict drinking outcomes following a brief intervention. Exp Clin
78. Carmack SA, Keeley RJ, Vendruscolo JCM, Lowery-Gionta EG, Lu H, Koob GF, Psychopharmacol. 2005;13:93–101.
et al. Heroin addiction engages negative emotional learning brain circuits in 105. Amlung MT, Acker J, Stojek MK, Murphy JG, MacKillop J. Is talk “cheap”? An
rats. J Clin Investig. 2019;129:2480–84. initial investigation of the equivalence of alcohol purchase task performance for
79. Spragg SDS. Morphine addiction in chimpanzees. Comp Psychol Monogr. hypothetical and actual rewards. Alcohol, Clin Exp Res. 2012;36:716–24.
1940;15:132–32. 106. Madden GJ, Bickel WK. Abstinence and price effects on demand for cigarettes: a
80. Negus SS. Choice between heroin and food in nondependent and heroin- behavioral-economic analysis. Addiction. 1999;94:577–88.
dependent rhesus monkeys: effects of naloxone, buprenorphine, and metha- 107. MacKillop J, Brown CL, Stojek MK, Murphy CM, Sweet L, Niaura RS. Behavioral
done. J Pharm Exp Ther. 2006;317:711–23. economic analysis of withdrawal- and cue-elicited craving for tobacco: an initial
81. Shaham Y, Stewart J. Stress reinstates heroin-seeking in drug-free animals: an investigation. Nicotine Tob Res. 2012;14:1426–34.
effect mimicking heroin, not withdrawal. Psychopharmacol (Berl). 108. Owens MM, Ray LA, MacKillop J. Behavioral economic analysis of stress effects
1995;119:334–41. on acute motivation for alcohol. J Exp Anal Behav. 2015;103:77–86.
82. Goldberg SR, Woods JH, Schuster CR. Morphine: conditioned increases in self- 109. Gray JC, MacKillop J. Interrelationships among individual differences in alcohol
administration in rhesus monkeys. Science. 1969;166:1306–07. demand, impulsivity, and alcohol misuse. Psychol Addict Behav. 2014;28:282–7.
83. Mantsch JR, Baker DA, Funk D, Le AD, Shaham Y. Stress-induced reinstatement 110. Murphy JG, Yurasek AM, Dennhardt AA, Skidmore JR, McDevitt-Murphy ME,
of drug seeking: 20 years of progress. Neuropsychopharmacology. MacKillop J, et al. Symptoms of depression and PTSD are associated with ele-
2016;41:335–56. vated alcohol demand. Drug Alcohol Depend. 2013;127:129–36.
84. Spanagel R, Noori HR, Heilig M. Stress and alcohol interactions: animal studies 111. Hogarth L, Chase HW. Parallel goal-directed and habitual control of human
and clinical significance. Trends Neurosci. 2014;37:219–27. drug-seeking: implications for dependence vulnerability. J Exp Psychol: Anim
85. Sommer WH, Rimondini R, Hansson AC, Hipskind PA, Gehlert DR, Barr CS, et al. Behav Process. 2011;37:261–76.
Upregulation of voluntary alcohol intake, behavioral sensitivity to stress, and 112. Hogarth L, Hardy L. Alcohol dependence symptoms are associated with greater
amygdala crhr1 expression following a history of dependence. Biol Psychiatry. relative value ascribed to alcohol, but not greater discounting of costs imposed
2008;63:139–45. on alcohol. Psychopharmacology. 2018;235:2257–66.
86. Mantsch JR, Baker DA, Francis DM, Katz ES, Hoks MA, Serge JP. Stressor- and 113. Hardy L, Hogarth L. A novel concurrent pictorial choice model of mood-induced
corticotropin releasing factor-induced reinstatement and active stress-related relapse in hazardous drinkers. Exp Clin Psychopharmacol. 2017;25:448–55.

Neuropsychopharmacology (2020) 0:1 – 16

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
13
114. Miele A, Thompson M, Jao NC, Kalhan R, Leone F, Hogarth L, et al. Cancer 139. Sinha R, Fox HC, Hong K, Hansen J, Tuit K, Kreek M. Effects of adrenal sensitivity,
patients enrolled in a smoking cessation clinical trial: characteristics and cor- stress- and cue-induced craving, and anxiety on subsequent alcohol relapse and
relates of smoking rate and nicotine dependence. J Addiction. 2018;2018:7. treatment outcomes. Arch Gen Psychiatry. 2011;68:942–52.
115. Hogarth L, Hardy L, Mathew AR, Hitsman B. Negative mood-induced alcohol- 140. Sinha R, Garcia M, Paliwal P, Kreek M, Rounsaville BJ. Stress-induced cocaine
seeking is greater in young adults who report depression symptoms, drinking to craving and hypothalamic-pituitary-adrenal responses are predictive of cocaine
cope, and subjective reactivity. Exp Clin Psychopharmacol. 2018;26:138–46. relapse outcomes. Arch Gen Psychiatry. 2006;63:324–31.
116. Moeller SJ, Beebe-Wang N, Woicik PA, Konova AB, Maloney T, Goldstein RZ. 141. Back SE, Hartwell K, DeSantis SM, Saladin M, McRae-Clark AL, Price KL, et al.
Choice to view cocaine images predicts concurrent and prospective drug use in Reactivity to laboratory stress provocation predicts relapse to cocaine. Drug
cocaine addiction. Drug Alcohol Depend. 2013;130:178–85. Alcohol Depend. 2010;106:21–27.
117. Moeller SJ, Maloney T, Parvaz MA, Dunning JP, Alia-Klein N, Woicik PA, et al. 142. Rousseau GS, Irons JG, Correia CJ. The reinforcing value of alcohol in a drinking
Enhanced choice for viewing cocaine pictures in cocaine addiction. Biol Psy- to cope paradigm. Drug Alcohol Depend. 2011;118:1–4.
chiatry. 2009;66:169–76. 143. Field M, Quigley M. Mild stress increases attentional bias in social drinkers who
118. Bickel WK, DeGrandpre RJ, Higgins ST, Hughes JR, Badger GJ. Effects of simu- drink to cope: a replication and extension. Exp Clin Psychopharmacol.
lated employment and recreation on drug taking: a behavioral economic ana- 2009;17:312–19.
lysis. Exp Clin Psychopharmacol. 1995;3:467–76. 144. Birch CD, Stewart SH, Wall A, McKee SA, Eisnor SJ, Theakston JA. Mood-induced
119. Hart CL, Haney M, Foltin RW, Fischman MW. Alternative reinforcers differentially increases in alcohol expectancy strength in internally motivated drinkers. Psy-
modify cocaine self-administration by humans. Behav Pharmacol. 2000;11:87–91. chol Addictive Behav. 2004;18:231–38.
120. Stoops WW, Lile JA, Glaser PEA, Hays LR, Rush CR. Alternative reinforcer 145. Grant VV, Stewart SH, Birch CD. Impact of positive and anxious mood on implicit
response cost impacts cocaine choice in humans. Prog Neuro-Psychopharmacol alcohol-related cognitions in internally motivated undergraduate drinkers.
Biol Psychiatry. 2012;36:189–93. Addict Behav. 2007;32:2226–37.
121. Hatsukami DK, Thompson TN, Pentel PR, Flygare BK, Carroll ME. Self- 146. Field M, Powell H. Stress increases attentional bias for alcohol cues in social
administration of smoked cocaine. Exp Clin Psychopharmacol. 1994;2:115–25. drinkers who drink to cope. Alcohol Alcohol. 2007;42:560–66.
122. Lile JA, Stoops WW, Rush CR, Negus SS, Glaser PEA, Hatton KW, et al. Devel- 147. Zack M, Poulos CX, Fragopoulos F, MacLeod CM. Effects of negative and positive
opment of a translational model to screen medications for cocaine use disorder mood phrases on priming of alcohol words in young drinkers with high and low
II: choice between intravenous cocaine and money in humans. Drug Alcohol anxiety sensitivity. Exp Clin Psychopharmacol. 2003;11:176–85.
Depend. 2016;165:111–19. 148. Austin JL, Smith JE. Drinking for negative reinforcement: the semantic priming
123. Comer SD, Metz VE, Cooper ZD, Kowalczyk WJ, Jones JD, Sullivan MA, et al. of alcohol concepts. Addict Behav. 2008;33:1572–80.
Comparison of a drug versus money and drug versus drug self-administration 149. Woud ML, Becker ES, Rinck M, Salemink E. The relationship between drinking
choice procedure with oxycodone and morphine in opioid addicts. Behav motives and alcohol-related interpretation biases. J Behav Ther Exp Psychiatry.
Pharmacol. 2013;24:504–16. 2015;47:102–10.
124. Cassidy RN, Tidey JW, Kahler CW, Wray TB, Colby SM. Increasing the value of an 150. Fucito LM, Juliano LM. Depression moderates smoking behavior in response to a
alternative monetary reinforcer reduces cigarette choice in adolescents. Nico- sad mood induction. Psychol Addictive Behav. 2009;23:546–51.
tine Tob Res. 2015;17:1449–55. 151. Sinha R, Fox HC, Hong KA, Bergquist K, Bhagwagar Z, Siedlarz KM. Enhanced
125. Hogarth L, Hardy L, Bakou AE, Mahlberg J, Weidemann G, Cashel S, et al. negative emotion and alcohol craving, and altered physiological responses
Negative mood induction increases choice of heroin versus food pictures in following stress and cue exposure in alcohol dependent individuals. Neu-
opiate dependent individuals: correlation with self-medication coping motives ropsychopharmacology. 2009;34:1198–208.
and subjective reactivity. Front Psychiatry. 2019;10:274. in press 152. Zack M, Poulos CX, Fragopoulos F, Woodford TM, MacLeod CM. Negative affect
126. Moeller SJ, Zilverstand A, Konova AB, Kundu P, Parvaz MA, Preston-Campbell R, words prime beer consumption in young drinkers. Addict Behav. 2006;31:
et al. Neural correlates of drug-biased choice in currently using and abstinent 169–73.
individuals with cocaine use disorder. Biol Psychiatry. 2017;3:485–94. in press 153. Randall DM, Cox WM. Experimental mood inductions in persons at high and low
127. Moeller SJ, Maloney T, Parvaz MA, Alia-Klein N, Woicik PA, Telang F, et al. risk for alcohol problems. Am J Drug Alcohol Abus. 2001;27:183–87.
Impaired insight in cocaine addiction: laboratory evidence and effects on 154. Zack M, Toneatto T, MacLeod CM. Implicit activation of alcohol concepts by
cocaine-seeking behaviour. Brain. 2010;133:1484–93. negative affective cues distinguishes between problem drinkers with high and
128. Chase HW, MacKillop J, Hogarth L. Isolating behavioural economic indices of low psychiatric distress. J Abnorm Psychol. 1999;108:518–31.
demand in relation to nicotine dependence. Psychopharmacology. 2013;226: 155. Boschloo L, Vogelzangs N, van den Brink W, Smit JH, Veltman DJ, Beekman AT,
371–80. et al. Depressive and anxiety disorders predicting first incidence of alcohol use
129. Perkins KA, Broge M, Gerlach D, Sanders M, Grobe JE, Cherry C, et al. Acute disorders: results of the Netherlands Study of Depression and Anxiety (NESDA). J
nicotine reinforcement, but not chronic tolerance, predicts withdrawal and Clin Psychiatry. 2013;74:1233–40.
relapse after quitting smoking. Health Psychol. 2002;21:332–39. 156. Kushner MG, Abrams K, Thuras P, Hanson KL, Brekke M, Sletten S. Follow-up
130. Hogarth L. Goal-directed and transfer-cue-elicited drug-seeking are dissociated study of anxiety disorder and alcohol dependence in comorbid alcoholism
by pharmacotherapy: evidence for independent additive controllers. J Exp treatment patients. Alcohol: Clin Exp Res. 2005;29:1432–43.
Psychol: Anim Behav Processes. 2012;38:266–78. 157. Bruce SE, Yonkers KA, Otto MW, Eisen JL, Weisberg RB, Pagano M, et al. Influence
131. Johnson MW, Bickel WK. The behavioral economics of cigarette smoking: the of psychiatric comorbidity on recovery and recurrence in generalized anxiety
concurrent presence of a substitute and an independent reinforcer. Behav disorder, social phobia, and panic disorder: a 12-year prospective study. Am J
Pharmacol. 2003;14:137–44. Psychiatry. 2005;162:1179–87.
132. Higgins ST, Roll JM, Bickel WK. Alcohol pretreatment increases preference for 158. King SM, Iacono WG, McGue M. Childhood externalizing and internalizing
cocaine over monetary reinforcement. Psychopharmacology. 1996;123:1–8. psychopathology in the prediction of early substance use. Addiction.
133. Vuchinich RE, Tucker JA. Behavioral theories of choice as a framework for 2004;99:1548–59.
studying drinking behaviour. J Abnorm Psychol. 1983;92:408–16. 159. Zimmermann P, Wittchen HU, Höfler M, Pfister H, Kessler RC, Lieb R. Primary
134. Vuchinich RE, Tucker JA, Rudd EJ. Preference for alcohol consumption as a anxiety disorders and the development of subsequent alcohol use disorders: a
function of amount and delay of alternative reward. J Abnorm Psychol. 4-year community study of adolescents and young adults. Psychol Med.
1987;96:259–63. 2003;33:1211–22.
135. Hogarth L, Field M, Rose AK. Phasic transition from goal-directed to habitual 160. Samet S, Fenton MC, Nunes E, Greenstein E, Aharonovich E, Hasin D. Effects of
control over drug-seeking produced by conflicting reinforcer expectancy. independent and substance-induced major depressive disorder on remission
Addict Biol. 2013;18:88–97. and relapse of alcohol, cocaine and heroin dependence. Addiction (Abingdon,
136. Brady KT, Back SE, Waldrop AE, McRae AL, Anton RF, Upadhyaya HP, et al. Cold Engl). 2013;108:115–23.
pressor task reactivity: predictors of alcohol use among alcohol-dependent 161. Sihvola E, Rose RJ, Dick DM, Pulkkinen L, Marttunen M, Kaprio J. Early-onset
individuals with and without comorbid posttraumatic stress disorder. Alcohol: depressive disorders predict the use of addictive substances in adolescence: a
Clin Exp Res. 2006;30:938–46. prospective study of adolescent Finnish twins. Addiction. 2008;103:2045–53.
137. Cooney NL, Litt MD, Morse PA, Bauer LO, Gaupp L. Alcohol cue reactivity, 162. Crum RM, Pratt LA. Risk of heavy drinking and alcohol use disorders in social
negative-mood reactivity, and relapse in treated alcoholic men. J Abnorm phobia: a prospective analysis. Am J Psychiatry. 2001;158:1693–700.
Psychol. 1997;106:243–50. 163. Crum RM, Mojtabai R, Lazareck S, Bolton JM, Robinson J, Sareen J, et al. A
138. Higley A, Crane N, Spadoni A, Quello S, Goodell V, Mason B. Craving in response prospective assessment of reports of drinking to self-medicate mood symptoms
to stress induction in a human laboratory paradigm predicts treatment outcome with the incidence and persistence of alcohol dependence. JAMA Psychiatry.
in alcohol-dependent individuals. Psychopharmacology. 2011;218:121–29. 2013;70:718–26.

Neuropsychopharmacology (2020) 0:1 – 16

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
14
164. Crum RM, La Flair L, Storr CL, Green KM, Stuart EA, Alvanzo AAH, et al. Reports of 192. Phillips GD, Vugler A. Effects of sensitization on the detection of an instrumental
drinking to self-medicate anxiety symptoms: longitudinal assessment for sub- contingency. Pharm Biochem Behav. 2011;100:48–58.
groups of individuals with alcohol dependence. Depress Anxiety. 193. Colwill RM, Rescorla RA. Instrumental responding remains sensitive to reinforcer
2013;30:174–83. devaluation after extensive training. J Exp Psychol Anim Behav Process.
165. Crum RM, Green KM, Storr CL, Chan YF, Ialongo N, Stuart EA, et al. Depressed 1985;11:520–36.
mood in childhood and subsequent alcohol use through adolescence and 194. Colwill RM, Triola SM. Instrumental responding remains under the control of the
young adulthood. Arch Gen Psychiatry. 2008;65:702–12. consequent outcome after extended training. Behav Process. 2002;57:51–64.
166. Gilman SE, Abraham HD. A longitudinal study of the order of onset of alcohol 195. Holland PC. Relations between Pavlovian-instrumental transfer and reinforcer
dependence and major depression. Drug Alcohol Depend. 2001;63:277–86. devaluation. J Exp Psychol Anim Behav Process. 2004;30:258–58.
167. Holahan CJ, Moos RH, Holahan CK, Cronkite RC, Randall PK. Drinking to cope, 196. Kosaki Y, Dickinson A. Choice and contingency in the development of beha-
emotional distress and alcohol use and abuse: a ten-year model. J Stud Alcohol. vioral autonomy during instrumental conditioning. J Exp Psychol: Anim Behav
2001;62:190–98. Process. 2010;36:334–42.
168. Vernig PM, Orsillo SM. Drinking motives and college alcohol problems: a pro- 197. Singer BF, Fadanelli M, Kawa AB, Robinson TE. Are cocaine-seeking “habits”
spective study. J Subst Use. 2015;20:340–46. necessary for the development of addiction-like behavior in rats? J Neurosci.
169. Hussong AM, Ennett ST, Cox MJ, Haroon M. A systematic review of the unique 2018;38:60.
prospective association of negative affect symptoms and adolescent substance 198. Heather N. Is the concept of compulsion useful in the explanation or description
use controlling for externalizing symptoms. Psychol Addictive Behav. of addictive behaviour and experience? Addictive Behav Rep. 2017;6:15–38.
2017;31:137–47. 199. Hogarth L, Chase HW, Baess K. Impaired goal-directed behavioural control in
170. Simpson TL, Miller WR. Concomitance between childhood sexual and physical human impulsivity. Q J Exp Psychol. 2012;65:305–16.
abuse and substance use problems: a review. Clin Psychol Rev. 2002;22:27–77. 200. Hogarth L, Lam-Cassettari C, Pacitti H, Currah T, Mahlberg J, Hartley L, et al.
171. Konkolÿ Thege B, Horwood L, Slater L, Tan MC, Hodgins DC, Wild TC. Rela- Intact goal-directed control in treatment-seeking drug users indexed by
tionship between interpersonal trauma exposure and addictive behaviors: a outcome-devaluation and Pavlovian to instrumental transfer: critique of habit
systematic review. BMC Psychiatry. 2017;17:164. theory. Eur J Neurosci. 2018;50:2513–25.
172. Butt S, Chou S, Browne K. A rapid systematic review on the association between 201. Luijten M, Franken IHA, Gillan CM, De Wit S, Robbins TW, Ersche KD. Goal-
childhood physical and sexual abuse and illicit drug use among males. Child directed and habitual control in smokers. Nicotine Tob Res. 2019;15:ntz001.
Abus Rev. 2011;20:6–38. 202. Sjoerds Z, de Wit S, van den Brink W, Robbins TW, Beekman AT, Penninx BW,
173. Brady KT, Back SE. Childhood trauma, posttraumatic stress disorder, and alcohol et al. Behavioral and neuroimaging evidence for overreliance on habit learning
dependence. Alcohol Res: Curr Rev. 2012;34:408–13. in alcohol-dependent patients. Transl Psychiatry. 2013;3:e337.
174. Miles FJ, Everitt BJ, Dickinson A. Oral cocaine seeking by rats: action or habit? 203. Ersche KD, Gillan CM, Jones PS, Williams GB, Ward LHE, Luijten M, et al. Carrots
Behav Neurosci. 2003;117:927–38. and sticks fail to change behavior in cocaine addiction. Science.
175. Mangieri RA, Cofresí RU, Gonzales RA. Ethanol seeking by long evans rats is not 2016;352:1468–71.
always a goal-directed behavior. PLoS ONE 2012;7:e42886. 204. Mitchell CJ, De Houwer J, Lovibond PF. The propositional nature of human
176. Loughlin A, Funk D, Coen K, Lê AD. Habitual nicotine-seeking in rats following associative learning. Behav Brain Sci. 2009;32:183–98.
limited training. Psychopharmacology. 2017;234:2619–29. 205. Lovibond PF. Cognitive processes in extinction. Learn Mem. 2004;11:495–500.
177. Schmitzer-Torbert N, Apostolidis S, Amoa R, O’Rear C, Kaster M, Stowers J, et al. 206. Newell BR, Lagnado DA, Shanks DR. Challenging the role of implicit processes in
Post-training cocaine administration facilitates habit learning and requires the probabilistic category learning. Psychonomic Bull Rev. 2007;14:505–11.
infralimbic cortex and dorsolateral striatum. Neurobiol Learn Mem. 207. Shanks DR, St. John MF. Characteristics of dissociable human learning systems.
2015;118:105–12. Behav Brain Sci. 1994;17:367–447.
178. Corbit LH, Chieng BC, Balleine BW. Effects of repeated cocaine exposure on 208. Hogarth L, Duka T. Human nicotine conditioning requires explicit contingency
habit learning and reversal by N-acetylcysteine. Neuropsychopharmacology. knowledge: is addictive behaviour cognitively mediated? Psychopharmacology.
2014;39:1893–901. 2006;184:553–66.
179. Nordquist RE, Voorn P, Malsen J, Joosten R, Pennartz CMA, Vanderschuren L. 209. De Houwer J, Tanaka A, Moors A, Tibboel H. Kicking the habit: why evidence for
Augmented reinforcer value and accelerated habit formation after repeated habits in humans might be overestimated. Motivation. Science. 2018;4:50–59.
amphetamine treatment. Eur Neuropsychopharmacol. 2007;17:532–40. 210. Potvin S, Stavro K, Rizkallah E, Pelletier J. Cocaine and cognition: a systematic
180. Nelson AJD, Killcross S. Accelerated habit formation following amphetamine quantitative review. J Addict Med. 2014;8:368–76.
exposure is reversed by D1, but enhanced by D2, receptor antagonists. Front 211. Stavro K, Pelletier J, Potvin S. Widespread and sustained cognitive deficits in
Neurosci. 2013;7:Article76. alcoholism: a meta-analysis. Addict Biol. 2013;18:203–13.
181. Nelson A, Killcross S. Amphetamine exposure enhances habit formation. J 212. Conti AA, McLean L, Tolomeo S, Steele JD, Baldacchino A. Chronic tobacco
Neurosci. 2006;26:3805–12. smoking and neuropsychological impairments: a systematic review and meta-
182. Schoenbaum G, Setlow B. Cocaine makes actions insensitive to outcomes but analysis. Neurosci Biobehav Rev. 2019;96:143–54.
not extinction: Implications for altered orbitofrontal-amygdalar function. Cereb 213. Gillan CM, Papmeyer M, Morein-Zamir S, Sahakian BJ, Fineberg NA, Robbins TW,
Cortex. 2005;15:1162–69. et al. Disruption in the balance between goal-directed behavior and habit
183. LeBlanc KH, Maidment NT, Ostlund SB. Repeated cocaine exposure facilitates learning in obsessive-compulsive disorder. Am J Psychiatry. 2011;168:718–26.
the expression of incentive motivation and induces habitual control in rats. PLoS 214. Pritchard T, Weidemann G, Hogarth L. Negative emotional appraisal selectively
ONE 2013;8:e61355. disrupts retrieval of expected outcome values required for goal-directed
184. Son J-H, Latimer C, Keefe KA. Impaired formation of stimulus–response, but not instrumental choice. Cognition Emot. 2018;32:843–51.
action–outcome, associations in rats with methamphetamine-induced neuro- 215. Fournier M, d’Arripe- Longueville F, Radel R. Effects of psychosocial stress on the
toxicity. Neuropsychopharmacology. 2011;36:2441–51. goal-directed and habit memory systems during learning and later execution.
185. Shiflett MW. The effects of amphetamine exposure on outcome-selective Pav- Psychoneuroendocrinology. 2017;77:275–83.
lovian-instrumental transfer in rats. Psychopharmacology. 2012;223:361–70. 216. Schwabe L, Wolf OT. Stress prompts habit behavior in humans. J Neurosci.
186. Ripley TL, Borlikova G, Lyons S, Stephens DN. Selective deficits in appetitive 2009;29:7191–98.
conditioning as a consequence of ethanol withdrawal. Eur J Neurosci. 217. Schwabe L, Wolf OT. Socially evaluated cold pressor stress after instrumental
2004;19:415–25. learning favors habits over goal-directed action. Psychoneuroendocrinology.
187. Dickinson A, Balleine BW, Watt A, Gonzalez F, Boakes RA. Motivational control 2010;35:977–86.
after extended instrumental training. Anim Learn Behav. 1995;23:197–206. 218. Schwabe L, Wolf OT. Stress-induced modulation of instrumental behavior: from
188. de Wit S, Kindt M, Knot SL, Verhoeven AAC, Robbins TW, Gasull-Camos J, et al. goal-directed to habitual control of action. Behav Brain Res. 2011;219:321–8.
Shifting the balance between goals and habits: five failures in experimental 219. Soares JM, Sampaio A, Ferreira LM, Santos NC, Marques F, Palha JA, et al. Stress-
habit induction. J Exp Psychol Gen. 2018;147:1043–65. induced changes in human decision-making are reversible. Transl Psychiatry.
189. Corbit LH, Nie H, Janak PH. Habitual responding for alcohol depends upon both 2012;2:e131.
AMPA and D2 receptor signaling in the dorsolateral striatum. Front Behav 220. Hogarth L, Attwood AS, Bate HA, Munafò MR. Acute alcohol impairs human
Neurosci. 2014;8:301. goal-directed action. Biol Psychol. 2012;90:154–60.
190. Perez OD, Dickinson A. A theory of actions and habits in free-operant behavior: 221. Furlong TM, Supit AS, Corbit LH, Killcross S, Balleine BW. Pulling habits out of
the interaction of rate correlation and contiguity systems. bioRxiv. 2019:807800. rats: adenosine 2A receptor antagonism in dorsomedial striatum rescues meth-
191. Trask S, Shipman ML, Green JT, Bouton ME. Some factors that restore goal- amphetamine-induced deficits in goal-directed action. Addict Biol. 2017;22:
direction to a habitual behavior. Neurobiol Learn Mem. 2020; in press. 172–83.

Neuropsychopharmacology (2020) 0:1 – 16

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
15
222. Ostlund SB, Maidment NT, Balleine BW. Alcohol-paired contextual cues produce 247. Lucantonio F, Stalnaker TA, Shaham Y, Niv Y, Schoenbaum G. The impact of
an immediate and selective loss of goal-directed action in rats. Front Integr orbitofrontal dysfunction on cocaine addiction. Nat Neurosci. 2012;15:358–66.
Neurosci. 2010;4:19. 248. Koob GF. Addiction is a reward deficit and stress surfeit disorder. Front Psy-
223. Chen J, Liang J, Lin X, Zhang Y, Zhang Y, Lu L, et al. Sleep deprivation promotes chiatry. 2013;4:72.
habitual control over goal-directed control: behavioral and neuroimaging evi- 249. Hogarth L, Martin L, Seedat S. Relationship between childhood abuse and sub-
dence. J Neurosci. 2017;37:11979. stance misuse problems is mediated by substance use coping motives, in school
224. Alvares GA, Balleine BW, Guastella AJ. Impairments in goal-directed actions attending South African adolescents. Drug Alcohol Depend. 2019;194:69–74.
predict treatment response to cognitive-behavioral therapy in social anxiety 250. Stewart SH, Sherry SB, Comeau MN, Mushquash CJ, Collins P, Van Wilgenburg H.
disorder. PLoS ONE. 2014;9:e94778. Hopelessness and excessive drinking among aboriginal adolescents: the med-
225. Alvares GA, Balleine BW, Whittle L, Guastella AJ. Reduced goal-directed action iating roles of depressive symptoms and drinking to cope. Depression Res Treat.
control in autism spectrum disorder. Autism Res. 2016;9:1285–93. 2011;2011:11.
226. Morris RW, Quail S, Griffiths KR, Green MJ, Balleine BW. Corticostriatal control of 251. Martin G, Inchley J, Currie C. Do drinking motives mediate the relationship
goal-directed action is impaired in schizophrenia. Biol Psychiatry. between neighborhood characteristics and alcohol use among adolescents? Int
2015;77:187–95. J Env Res Public Health. 2019;16:853.
227. de Wit S, Barker RA, Dickinson AD, Cools R. Habitual versus goal-directed action 252. Mohr CD, McCabe CT, Haverly SN, Hammer LB, Carlson KF. Drinking motives and
control in Parkinson disease. J Cogn Neurosci. 2011;23:1218–29. alcohol use: the SERVe study of U.S. Current and former service members. J Stud
228. Klossek UMH, Russell J, Dickinson A. The control of instrumental action following Alcohol Drugs. 2018;79:79–87.
outcome devaluation in young children aged between 1 and 4 years. J Exp 253. Miller SM, Pedersen ER, Marshall GN. Combat experience and problem drinking
Psychol Gen. 2008;137:39–51. in veterans: exploring the roles of PTSD, coping motives, and perceived stigma.
229. Sebold M, Deserno L, Nebe S, Schad DJ, Garbusow M, Hagele C, et al. Model- Addict Behav. 2017;66:90–95.
based and model-free decisions in alcohol dependence. Neuropsychobiology. 254. Øverup CS, DiBello AM, Brunson JA, Acitelli LK, Neighbors C. Drowning the pain:
2014;70:122–31. intimate partner violence and drinking to cope prospectively predict problem
230. Patzelt EH, Kool W, Millner AJ, Gershman SJ. Incentives boost model-based drinking. Addict Behav. 2015;41:152–61.
control across a range of severity on several psychiatric constructs. Biol Psy- 255. Vilhena-Churchill N, Goldstein AL. Child maltreatment and marijuana problems
chiatry. 2019;85:425–33. in young adults: examining the role of motives and emotion dysregulation.
231. Otto AR, Gershman SJ, Markman AB, Daw ND. The curse of planning: Dissecting Child Abus Negl. 2014;38:962–72.
multiple reinforcement-learning systems by taxing the central executive. Psy- 256. Chinneck A, Thompson K, Dobson KS, Stuart H, Teehan M, Stewart SH. Neurotic
chol Sci. 2013;24:751–61. personality traits and risk for adverse alcohol outcomes: chained mediation
232. Otto AR, Raio CM, Chiang A, Phelps EA, Daw ND. Working-memory capacity through emotional disorder symptoms and drinking to cope. Subst Use Misuse.
protects model-based learning from stress. Proc Natl Acad Sci USA. 2018;53:1730–41.
2013;110:201941 257. Collins JL, Thompson K, Sherry SB, Glowacka M, Stewart SH. Drinking to cope
233. Reiter AMF, Deserno L, Wilbertz T, Heinze H-J, Schlagenhauf F. Risk factors for with depression mediates the relationship between social avoidance and
addiction and their association with model-based behavioral control. Front alcohol problems: a 3-wave, 18-month longitudinal study. Addict Behav.
Behav Neurosci. 2016;10:26. 2018;76:182–87.
234. Schad DJ, Jünger E, Sebold M, Garbusow M, Bernhardt N, Javadi A-H, et al. 258. Dvorak RD, Pearson MR, Day AM. Ecological momentary assessment of acute
Processing speed enhances model-based over model-free reinforcement alcohol use disorder symptoms: associations with mood, motives, and use on
learning in the presence of high working memory functioning. Front Psychol. planned drinking days. Exp Clin Psychopharmacol. 2014;22:285–97.
2014;5:1450–50. 259. Asberg K, Renk K. Substance use coping as a mediator of the relationship
235. MacKillop J, Murphy JG. A behavioral economic measure of demand for alcohol between trauma symptoms and substance use consequences among incar-
predicts brief intervention outcomes. Drug Alcohol Depend. 2007;89:227–33. cerated females with childhood sexual abuse histories. Subst Use Misuse.
236. Murphy JG, MacKillop J. Relative reinforcing efficacy of alcohol among college 2012;47:799–808.
student drinkers. Exp Clin Psychopharmacol. 2006;14:219–27. 260. O’Hare T, Sherrer M. Drinking motives as mediators between PTSD symptom
237. MacKillop J, Miranda R, Monti PM, Ray LA, Murphy JG, Rohsenow DJ, et al. severity and alcohol consumption in persons with severe mental illnesses.
Alcohol demand, delayed reward discounting, and craving in relation to Addict Behav. 2011;36:465–69.
drinking and alcohol use disorders. J Abnorm Psychol. 2010;119:106–14. 261. Topper LR, Castellanos-Ryan N, Mackie C, Conrod PJ. Adolescent bullying vic-
238. MacKillop J, Jackson JN, Murphy JG, Amlung MT. Associations between indivi- timisation and alcohol-related problem behaviour mediated by coping drinking
dual differences in alcohol’s relative reinforcing value and alcohol misuse: a motives over a 12 month period. Addict Behav. 2011;36:6–13.
meta-analysis. Alcoholism: Clinical and Experimental Research. 2015;39. 262. Fossos N, Kaysen D, Neighbors C, Lindgren KP, Hove MC. Coping motives as a
239. Zvorsky I, Nighbor TD, Kurti AN, DeSarno M, Naudé G, Reed DD, et al. Sensitivity mediator of the relationship between sexual coercion and problem drinking in
of hypothetical purchase task indices when studying substance use: A sys- college students. Addict Behav. 2011;36:1001–07.
tematic literature review. Preventive Med. 2019;128:105789. 263. Gonzalez VM, Reynolds B, Skewes MC. Role of impulsivity in the relationship
240. Gonzalez-Roz A, Jackson J, Murphy C, Rohsenow DJ, MacKillop J. Behavioral between depression and alcohol problems among emerging adult college
economic tobacco demand in relation to cigarette consumption and nicotine drinkers. Exp Clin Psychopharmacol. 2011;19:303–13.
dependence: a meta-analysis of cross-sectional relationships. Addiction . 264. Goldstein AL, Flett GL, Wekerle C. Child maltreatment, alcohol use and drinking
2019;114:1926–40. consequences among male and female college students: an examination of
241. González-Roz A, Secades-Villa R, Weidberg S, García-Pérez Á, Reed DD. Latent drinking motives as mediators. Addict Behav. 2010;35:636–39.
structure of the cigarette purchase task among treatment-seeking smokers with 265. Yeater EA, Austin JL, Green MJ, Smith JE. Coping mediates the relationship
depression and its predictive validity on smoking abstinence. Nicotine Tob Res. between posttraumatic stress disorder (PTSD) symptoms and alcohol use in
2018. homeless, ethnically diverse women: a preliminary study. Psychol Trauma:
242. Murphy JG, Yurasek AM, Meshesha LZ, Dennhardt AA, Mackillop J, Skidmore JR, Theory, Res, Pract, Policy. 2010;2:307–10.
et al. Family history of problem drinking is associated with less sensitivity of 266. Gonzalez VM, Bradizza CM, Collins RL. Drinking to cope as a statistical mediator
alcohol demand to a next-day responsibility. J Stud Alcohol Drugs. in the relationship between suicidal ideation and alcohol outcomes among
2014;75:653–63. underage college drinkers. Psychol Addictive Behav. 2009;23:443–51.
243. Fortier CB, LaFleche G, Disterhoft JF, Steffen EM, Venne JR, McGlinchey RE. Delay 267. Young-Wolff KC, Kendler KS, Sintov ND, Prescott CA. Mood-related drinking
discrimination and reversal eyeblink classical conditioning in abstinent chronic motives mediate the familial association between major depression and alcohol
alcoholics. Neuropsychology. 2008;22:196–208. dependence. Alcohol: Clin Exp Res. 2009;33:1476–86.
244. Reiter AMF, Deserno L, Kallert T, Heinze H-J, Heinz A, Schlagenhauf F. Behavioral 268. Mooney JL, Minor KI, Wells JB, Leukefeld C, Oser CB, Staton Tindall M. The
and neural signatures of reduced updating of alternative options in alcohol- relationship of stress, impulsivity, and beliefs to drug use severity in a sample of
dependent patients during flexible decision-making. J Neurosci. 2016;36:10935. women prison inmates. Int J Offender Ther Comp Criminol. 2008;52:686–97.
245. Vanes LD, van Holst RJ, Jansen JM, van den Brink W, Oosterlaan J, Goudriaan AE. 269. Kaysen D, Dillworth TM, Simpson T, Waldrop A, Larimer ME, Resick PA. Domestic
Contingency learning in alcohol dependence and pathological gambling: learning violence and alcohol use: trauma-related symptoms and motives for drinking.
and unlearning reward contingencies. Alcohol: Clin Exp Res. 2014;38:1602–10. Addict Behav. 2007;32:1272–83.
246. Jentsch JD, Olausson P, De La Garza R II, Taylor JR. Impairments of reversal 270. Schuckit MA, Smith TL, Chacko Y. Evaluation of a depression-related model of
learning and response perseveration after repeated, intermittent cocaine alcohol problems in 430 probands from the San Diego prospective study. Drug
administrations to monkeys. Neuropsychopharmacology. 2002;26:183–90. Alcohol Depend. 2006;82:194–203.

Neuropsychopharmacology (2020) 0:1 – 16

 Addiction is driven by excessive goal-directed drug choice under negative. . .
L Hogarth
16
271. Ullman SE, Filipas HH, Townsend SM, Starzynski LL. Trauma exposure, post- 290. Pelloux Y, Murray JE, Everitt BJ. Differential vulnerability to the punishment of
traumatic stress disorder and problem drinking in sexual assault survivors. J cocaine related behaviours: effects of locus of punishment, cocaine taking
Stud Alcohol. 2005;66:610–19. history and alternative reinforcer availability. Psychopharmacology.
272. Grayson CE, Nolen-Hoeksema S. Motives to drink as mediators between child- 2015;232:125–34.
hood sexual assault and alcohol problems in adult women. J Traum. Stress. 291. James MH, Bowrey HE, Stopper CM, Aston-Jones G. Demand elasticity predicts
2005;18:137–45. addiction endophenotypes and the therapeutic efficacy of an orexin/hypocre-
273. Reardon ML, Lang AR, Patrick CJ. An evaluation of relations among antisocial tin-1 receptor antagonist in rats. Eur J Neurosci. 2018;50:2602–12.
behavior, psychopathic traits, and alcohol problems in incarcerated men. 292. Deroche-Gamonet V, Belin D, Piazza PV. Evidence for addiction-like behavior in
Alcohol: Clin Exp Res. 2002;26:1188–97. the rat. Science. 2004;305:1014–17.
274. Stewart SH, Zvolensky MJ, Eifert GH. Negative-reinforcement drinking motives 293. Kasanetz F, Deroche-Gamonet V, Berson N, Balado E, Lafourcade M, Manzoni O,
mediate the relation between anxiety sensitivity and increased drinking beha- et al. Transition to addiction is associated with a persistent impairment in
vior. Pers Individ Differ. 2001;31:157–71. synaptic plasticity. Science. 2010;328:1709.
275. Schuck AM, Widom CS. Childhood victimization and alcohol symptoms in 294. Vanderschuren LJMJ, Everitt BJ. Drug seeking becomes compulsive after pro-
females: causal inferences and hypothesized mediators. Child Abus Negl. longed cocaine self-administration. Science. 2004;305:1017–19.
2001;25:1069–92. 295. Augier E, Barbier E, Dulman RS, Licheri V, Augier G, Domi E, et al. A molecular
276. Peirce RS, Frone MR, Russell M, Cooper ML. Relationship of financial strain and mechanism for choosing alcohol over an alternative reward. Science.
psychosocial resources to alcohol use and abuse: the mediating role of negative 2018;360:1321–26.
affect and drinking motives. J Health Soc Behav. 1994;35:291–308. 296. Spanagel R. Animal models of addiction. Dialogues Clin Neurosci.
277. Schlagintweit HE, Thompson K, Goldstein AL, Stewart SH. An investigation of the 2017;19:247–58.
association between shame and problem gambling: the mediating role of 297. Muller CP, Schumann G. Drugs as instruments: a new framework for non-
maladaptive coping motives. J Gambl Stud. 2017;33:1067–79. addictive psychoactive drug use. Behav Brain Sci. 2011;34:293–310.
278. Voon V, Derbyshire K, Rück C, Irvine MA, Worbe Y, Enander J, et al. Disorders of 298. Hogarth L, Troisi JRI. A hierarchical instrumental decision theory of nicotine
compulsivity: a common bias towards learning habits. Mol Psychiatry. dependence. In: Balfour DJK, Munafò MR, editors. The neurobiology and
2015;20:345–52. genetics of nicotine and tobacco. Cham: Springer International Publishing; 2015.
279. Sebold M, Nebe S, Garbusow M, Guggenmos M, Schad DJ, Beck A, et al. When p. 165−91.
habits are dangerous: alcohol expectancies and habitual decision making pre- 299. MacKillop J. The behavioral economics and neuroeconomics of alcohol use
dict relapse in alcohol dependence. Biol Psychiatry. 2017;82:847–56. disorders. Alcohol: Clin Exp Res. 2016;40:672–85.
280. Nebe S, Kroemer NB, Schad DJ, Bernhardt N, Sebold M, Müller DK, et al. No 300. Berkman ET, Hutcherson CA, Livingston JL, Kahn LE, Inzlicht M. Self-control as
association of goal-directed and habitual control with alcohol consumption in value-based choice. Curr Dir Psychol Sci. 2017;26:422–28.
young adults. Addict Biol. 2018;23:379–93. 301. Shuai R, Bakou AE, Hardy L, Hogarth L. Ultra-brief breath counting (mindfulness)
281. Byrne KA, Otto AR, Pang B, Patrick CJ, Worthy DA. Substance use is associated training promotes recovery from stress-induced alcohol-seeking in student
with reduced devaluation sensitivity. Cogn Affect Behav Neurosci. drinkers. Addict Behav. 2020;102:106141.
2019;19:40–55. 302. Hogarth L, Hardy L, Bakou A, Mahlberg J, Weidemann G, Cashel S, et al. Negative
282. Gillan CM, Kosinski M, Whelan R, Phelps EA, Daw ND. Characterizing a psy- mood induction increases choice of heroin versus food pictures in opiate-
chiatric symptom dimension related to deficits in goal-directed control. eLife. dependent individuals: correlation with self-medication coping motives and
2016;5:e11305. subjective reactivity. Front Psychiatry. 2019;10:274.
283. Belin D, Mar AC, Dalley JW, Robbins TW, Everitt BJ. High impulsivity predicts the 303. Hogarth L. A critical review of habit theory of drug dependence. In: Verplanken
switch to compulsive cocaine-taking. Science. 2008;320:1352–55. B, editor. The psychology of habit: theory, mechanisms, change, and contexts.
284. Jonkman S, Pelloux Y, Everitt BJ. Drug intake is sufficient, but conditioning is not Cham: Springer; 2018. .
necessary for the emergence of compulsive cocaine seeking after extended self-
administration. Neuropsychopharmacology. 2012;37:1612–9.
285. Datta U, Martini M, Fan M, Sun W. Compulsive sucrose- and cocaine-seeking Open Access This article is licensed under a Creative Commons
behaviors in male and female Wistar rats. Psychopharmacology (Berlin). Attribution 4.0 International License, which permits use, sharing,
2018;235:2395–405. adaptation, distribution and reproduction in any medium or format, as long as you give
286. Datta U, Martini M, Sun W. Different functional domains measured by cocaine appropriate credit to the original author(s) and the source, provide a link to the Creative
self-administration under the progressive-ratio and punishment schedules in Commons license, and indicate if changes were made. The images or other third party
male Wistar rats. Psychopharmacology (Berlin). 2018;235:897–907. material in this article are included in the article’s Creative Commons license, unless
287. Pelloux Y, Everitt BJ, Dickinson A. Compulsive drug seeking by rats under pun- indicated otherwise in a credit line to the material. If material is not included in the
ishment: effects of drug taking history. Psychopharmacology. 2007;194:127–37. article’s Creative Commons license and your intended use is not permitted by statutory
288. Belin D, Berson N, Balado E, Piazza PV, Deroche-Gamonet V. High-novelty- regulation or exceeds the permitted use, you will need to obtain permission directly
preference rats are predisposed to compulsive cocaine self-administration. from the copyright holder. To view a copy of this license, visit http://creativecommons.
Neuropsychopharmacology. 2010;36:569. org/licenses/by/4.0/.
289. Bentzley BS, Jhou TC, Aston-Jones G. Economic demand predicts addiction-like
behavior and therapeutic efficacy of oxytocin in the rat. Proc Natl Acad Sci USA.
2014;111:11822–27. © The Author(s) 2020

Neuropsychopharmacology (2020) 0:1 – 16