Craving

Abstract

Craving is a type of drive: a strong desire for a drug, a food or another reward, and since DSM-5 a diagnostic criterion for substance use disorders. This article covers how craving is defined and why the term travels poorly across languages, the conditioning, cognitive, incentive-sensitization and elaborated-intrusion accounts of how it arises, cue reactivity, the meta-analytic link between craving and later use or relapse, the incubation of craving during abstinence, food craving, and how craving can be regulated. The central question is how much craving explains: drug cues and craving prospectively predict use and relapse, yet the assumption that craving is the underlying basis of addiction has been challenged.

Keywords: craving, cue reactivity, relapse

Craving is a central feature of addiction. Its inclusion as a diagnostic criterion for substance use disorders in DSM-5 came at a time when craving research was at an all-time high (Sayette, 2016), and the construct now spans drugs, foods (Pelchat, 2002) and, in one recent study, social contact (Tomova et al., 2020).

What Craving Is

Craving is hard to define. It is a term commonly used by North American lay people and an important category in psychological and addiction research, but difficulties in defining craving suggest that it may not be a natural category (Hormes & Rozin, 2010). Based on native speaker informants, only 17% of 20 languages lexicalize craving, and where a synonym exists it seems to mean a desire for a potential ingestant or a drug (Hormes & Rozin, 2010). In addiction research craving is described as a strong desire for drugs and a central feature of addiction (Vafaie & Kober, 2022; Sayette, 2016).

The same term covers food. A craving for specific foods represents one of the most common and intense experiences surrounding eating (Weingarten & Elston, 1990), and one review asks whether it is more than a linguistic accident that craving is used to describe intense desires for both foods and drugs of abuse (Pelchat, 2002). Craving has also entered diagnosis: the DSM-5 work group recommended combining abuse and dependence criteria into a single substance use disorder, dropping legal problems and adding craving as a criterion (Hasin et al., 2013). Measuring it is a problem in its own right: because there is no single perfect measure of craving, a review of its measurement argued that researchers must understand the limitations of each measure, and reviewed non-verbal methods alongside self-report (Sayette et al., 2000). A review of its clinical significance concluded that craving has considerable utility for diagnosis and as a clinical outcome (Tiffany & Wray, 2012).

MeSH files craving as a narrower descriptor of drive and lists no narrower descriptors beneath it.

Theories of Craving

Theories of craving are many. A review of 18 models spanning roughly 60 years (1948-2009) sorts them into conditioning-based, cognitive, psychobiological and motivation models, and states that it is well accepted that no single model explains craving completely (Skinner & Aubin, 2010). An earlier review classified the main models as phenomenological, conditioning and cognitive, and likewise concluded that no one specific theory provides a complete explanation of the phenomenon of craving (Drummond, 2001).

Conditioning models treat craving as learned. In them, craving is assumed to be an automatic, unconscious reaction to a stimulus (Skinner & Aubin, 2010); one influential analysis examined the role of unconditioned and conditioned drug effects in the self-administration of opiates and stimulants (Stewart et al., 1984), and the cue-reactivity meta-analysis described below set out to determine which, if any, learning-based model of cue reactivity the findings best support (Carter & Tiffany, 1999).

The cognitive model challenged the assumption that urges drive drug use. It argued that available data do not strongly support the assumptions that urges are necessary for drug use and responsible for relapse, and proposed instead that drug use in the addict is controlled by automatized action schemata, with urges conceptualized as responses supported by nonautomatic cognitive processes activated in parallel (Tiffany, 1990).

The incentive-sensitization theory asked why addicts crave drugs and why craving persists after long abstinence. It posits that addictive drugs enhance mesotelencephalic dopamine neurotransmission, that this system attributes incentive salience to events associated with its activation, and that in some individuals repeated drug use sensitizes it, so that it is specifically the sensitization of incentive salience that transforms ordinary 'wanting' into excessive drug craving (Robinson & Berridge, 1993). On this theory, sensitization of 'wanting' can occur independently of changes in drug 'liking' and in the systems that mediate withdrawal (Robinson & Berridge, 1993). A negative-reinforcement reformulation offers a rival motive: it proposes that the escape and avoidance of negative affect is the prepotent motive for addictive drug use, that negative affect is the motivational core of the withdrawal syndrome, and that addicted organisms learn to detect interoceptive cues of negative affect preconsciously (Baker et al., 2004).

The elaborated intrusion theory of desire places craving in conscious cognition. Intrusive thoughts about appetitive targets are triggered automatically by external or physiological cues and by cognitive associates; when intrusions elicit significant pleasure or relief, cognitive elaboration usually ensues and competes with concurrent cognitive tasks through retention of target-related information in working memory (Kavanagh et al., 2005). Desire images are momentarily rewarding but amplify awareness of somatic and emotional deficits (Kavanagh et al., 2005). The theory was later applied to food craving (May et al., 2012).

Cue Reactivity

The cue-reactivity procedure exposes addicts to drug-related stimuli while self-reported craving and physiological responses are monitored (Carter & Tiffany, 1999). A meta-analysis of 41 such studies found an effect size for craving of +0.92 across all addict groups; alcoholics had a significantly smaller craving effect size (+0.53) than other groups (+1.18 to +1.29), and physiological responses were smaller, with increased heart rate (+0.26) and sweat gland activity (+0.40) and decreased skin temperature (-0.24) (Carter & Tiffany, 1999). Imaging located part of the response: in detoxified cocaine users, a cocaine video elicited craving together with increases in limbic (amygdala and anterior cingulate) cerebral blood flow and decreases in basal ganglia blood flow, a pattern absent in cocaine-naive comparison subjects, and the authors concluded that limbic activation is one component of cue-induced cocaine craving (Childress et al., 1999). Dopamine was later linked to the same response: in eighteen cocaine-addicted subjects, a cocaine-cue video reduced the specific binding of a dopamine-sensitive radioligand in the dorsal but not the ventral striatum, and the magnitude of this reduction correlated with self-reports of craving (Volkow et al., 2006).

Figure 1

Effect Sizes for Responses to Drug Cues Versus Neutral Cues

Food cues have been studied in the same way. According to learning-based models, food cue reactivity and craving are conditioned responses that lead to increased eating and weight gain, and a meta-analysis of 69 statistics from 45 reports representing 3,292 participants found an overall medium effect of food cue reactivity and craving on eating and weight-related outcomes (r = 0.33, approximately 11% of variance) (Boswell & Kober, 2016). Visual food cues were associated with a similar effect size to real food exposure and a stronger effect size than olfactory cues (Boswell & Kober, 2016).

Craving, Use and Relapse

Whether craving predicts relapse has been contested. Despite decades of research, the roles of drug cues and craving in drug use and relapse remain controversial; a meta-analysis of 237 studies representing 51,788 participants found a significant prospective association of cue and craving indicators with drug use or relapse (OR 2.05, 95% CI 1.94-2.15), such that a 1-unit increase in cue and craving indicators was associated with more than double the odds of future drug use or relapse (Vafaie & Kober, 2022). Trim-and-fill analysis brought the adjusted effect size to an odds ratio of 1.31, and some of the strongest associations were found for cue-induced craving and same-day time lags (Vafaie & Kober, 2022).

Earlier reviews were more cautious. One concluded that theories that do not require craving to be present for relapse to occur have more empirical support than those that provide simplistic causal explanations, and challenged the relatively uncritical assumption that craving is the underlying basis of addiction and the most appropriate target for treatment (Drummond, 2001). Craving can also grow with abstinence: in rats, cocaine craving was reported to incubate after withdrawal (Grimm et al., 2001).

Food Craving

Food cravings raise the question of shared mechanisms. One review argues that many lines of evidence support a common mechanism for food and drug cravings in humans: opioid peptides seem to influence food palatability, there is mounting evidence for comorbidity between drug or alcohol abuse and excessive craving or liking for sweets, and, anecdotally, elderly individuals tend to 'age-out' of drug abuse and also experience markedly fewer food cravings with age (Pelchat, 2002). A review of the phenomenology of food craving suggested that progress requires addressing the definition and measurement of the craving state and recognizing the inadequacies of the ubiquitous assumption that cravings serve to identify and redress bodily needs (Weingarten & Elston, 1990). The topic is covered further under food addiction.

Regulating Craving

Craving can be regulated cognitively. In cigarette smokers using cognitive strategies to regulate craving for cigarettes and food, down-regulation was associated with activity in dorsomedial, dorsolateral and ventrolateral prefrontal cortices and with decreased activity in regions previously associated with craving, including the ventral striatum, subgenual cingulate, amygdala and ventral tegmental area; ventral striatal activity fully mediated the relationship between lateral prefrontal cortex and reported craving (Kober et al., 2010).

Mindfulness is a second route. A review of 30 experimental studies found most support for the elaborated intrusion theory of desire and conditioning models, and suggested that immediate reductions in craving from mindfulness strategies are likely to stem from working memory load and will not necessarily be superior to alternative strategies that also load working memory (Tapper, 2018). A smaller number of studies of mindfulness practised repeatedly over a longer period showed promising results (Tapper, 2018).

Worked Example

The relapse meta-analysis can be read two ways. The pooled odds ratio of 2.05 (Vafaie & Kober, 2022) means more than double the odds of later use or relapse per unit of cue or craving indicator, while the trim-and-fill estimate of 1.31 means about 31% higher odds. Taken at face value, the trim-and-fill adjustment removes roughly 70% of the excess odds, (1.05 - 0.31) / 1.05, although the authors note that trim and fill's interpretive utility is limited in the presence of high variability (Vafaie & Kober, 2022). For food, a correlation of r = 0.33 (Boswell & Kober, 2016) corresponds to 0.33 squared, about 0.109, the approximately 11% of variance the authors report.

Discussion

Three conclusions follow for cognitive psychology. First, craving is a measurable construct with uncertain boundaries. Drug cues reliably raise self-reported craving (Carter & Tiffany, 1999), but the term may not be a natural category, and, by native speaker informants, only 17% of 20 languages lexicalize it (Hormes & Rozin, 2010).

Second, craving predicts behavior, but not alone. Cue and craving indicators prospectively predict drug use and relapse (Vafaie & Kober, 2022), yet a cognitive model holds that drug use is controlled by automatized action schemata, with urges running in parallel (Tiffany, 1990), and a review found more support for theories that do not require craving to be present for relapse (Drummond, 2001).

Third, no single theory is complete. Reviews agree that no single model explains craving completely (Skinner & Aubin, 2010; Drummond, 2001): incentive sensitization explains why 'wanting' can grow while 'liking' does not (Robinson & Berridge, 1993), and elaborated intrusion theory explains why craving competes with working memory (Kavanagh et al., 2005).

Current Directions

Three lines of work are active. Diagnosis: craving was added to DSM-5 on the work group's recommendation (Hasin et al., 2013), and a meta-analysis concluded that its results support incorporating craving assessment across stages of treatment, as early as primary care (Vafaie & Kober, 2022). Social craving: after 10 hours of total social isolation, participants felt lonely and craved social interaction, and midbrain regions showed selective activation to social cues after isolation and to food cues after fasting, correlated with self-reported craving (Tomova et al., 2020). Laboratory method: a critique of laboratory craving studies raised questions about the conceptual and methodological assumptions underlying craving-induction paradigms and about the insight individuals have into their own future cravings (Sayette, 2016).

Common Misconceptions

“Craving is a universal human experience with a word in every language.”

Based on native speaker informants, only 17% of 20 languages lexicalize craving (Hormes & Rozin, 2010).

“Craving is the cause of addiction.”

A review challenged the relatively uncritical assumption that craving is the underlying basis of addiction and found more support for theories that do not require craving to be present for relapse (Drummond, 2001), although a later meta-analysis concluded that drug cue and craving indicators are an important mechanism underlying substance use disorders (Vafaie & Kober, 2022).

“Food cravings signal what the body needs.”

A review of food craving recognized the inadequacies of the ubiquitous assumption that cravings serve to identify and redress bodily needs (Weingarten & Elston, 1990).

Glossary

Automatized action schemata.

In the cognitive model of drug urges, the automatic processes held to control drug use in the addict.

Cognitive elaboration.

In elaborated intrusion theory, the processing that usually follows an intrusion that elicits significant pleasure or relief.

Craving.

A strong desire for a drug or another reward, and a DSM-5 diagnostic criterion for substance use disorders.

Cue reactivity.

The craving and physiological responses shown when people are exposed to drug-related or food stimuli.

Cue-induced craving.

Craving elicited by exposure to drug or food cues; for drugs, one of the indicators most strongly associated with later use or relapse.

Desire images.

Sensory images of an appetitive target, momentarily rewarding but amplifying awareness of deficits.

Elaborated intrusion theory.

The account of desire as intrusive thoughts about appetitive targets that are elaborated in working memory.

Food craving.

A craving for a specific food, one of the most common and intense experiences surrounding eating.

Incentive salience.

The attribution that makes stimuli attractive and 'wanted', which incentive-sensitization theory holds is sensitized in addiction.

Incentive-sensitization theory.

The theory that sensitization of incentive salience transforms ordinary 'wanting' into excessive drug craving.

Incubation of craving.

The reported growth of drug craving over the course of withdrawal.

Relapse.

A return to drug use after abstinence, prospectively associated with cue and craving indicators.

Tonic craving.

Craving measured without cue exposure, which also predicted eating behavior in meta-analysis.

Trim-and-fill analysis.

A meta-analytic adjustment for possible publication bias, which reduced the craving-relapse odds ratio from 2.05 to 1.31.

Key Researchers

Kent C. Berridge

. Professor in the Department of Psychology at the University of Michigan; co-authored the incentive-sensitization theory of drug craving. ORCID - Wikipedia - Faculty Page

Terry E. Robinson

. Professor at the University of Michigan; co-authored the incentive-sensitization theory of drug craving. ORCID - Wikipedia - Faculty Page

Michael A. Sayette

. Distinguished Professor in the Department of Psychology at the University of Pittsburgh; reviewed the theoretical and methodological issues in laboratory craving research. ORCID - Faculty Page

Frequently Asked Questions

What is craving in psychology?

Craving is a strong desire for drugs and a central feature of addiction (Vafaie & Kober, 2022; Sayette, 2016); the same term is used for intense desires for foods (Pelchat, 2002).

Is craving part of the diagnosis of addiction?

Yes: the DSM-5 work group recommended adding craving as a criterion for substance use disorder (Hasin et al., 2013).

Does craving predict relapse?

A meta-analysis of 237 studies found that cue and craving indicators were associated with more than double the odds of future drug use or relapse, an odds ratio that fell to 1.31 after trim-and-fill adjustment (Vafaie & Kober, 2022).

What is cue reactivity?

It is the response to drug-related stimuli measured in the cue-reactivity procedure, in which self-reported craving rose with an effect size of +0.92 across addict groups (Carter & Tiffany, 1999).

Why does craving persist after long abstinence?

Incentive-sensitization theory proposes that repeated drug use can sensitize the dopamine systems that attribute incentive salience, perhaps permanently (Robinson & Berridge, 1993), and in rats cocaine craving was reported to incubate after withdrawal (Grimm et al., 2001).

Are food cravings like drug cravings?

One review argues that many lines of evidence support common mechanisms for food and drug cravings (Pelchat, 2002), and food cue reactivity and craving predicted eating and weight gain with a medium effect (Boswell & Kober, 2016).

Can craving be controlled?

Cognitive down-regulation of craving engaged prefrontal regions and reduced ventral striatal activity (Kober et al., 2010), and mindfulness strategies may reduce craving in the short term, likely through working memory load (Tapper, 2018).

Can people crave social contact?

After 10 hours of social isolation, participants craved social interaction, and midbrain responses to social cues correlated with self-reported craving (Tomova et al., 2020).

References