Substance use disorder (SUD), or addiction, is a psychiatric disorder characterized by the loss of control over substance use, leading to persistent use despite related negative consequences and a prioritization of substance use over alternative rewards and meaningful activities (American Psychiatric Association, 2013; Auriacombe et al., 2018). In diagnostic classifications, the extension of the concept of addiction to rewarding behaviors void of substance intake has progressively evolved. In the DSM-IV, pathological gambling was listed among “Impulse-Control Disorders.” In 2013, the DSM-5 opened the “Substance-Related and Addictive Disorders” category to both substance and behavioral addictions, including pathological gambling (Petry et al., 2013, 2014), whereas Internet Gaming Disorder (IGD) was included as a disorder for further study (American Psychiatric Association, 2013). In parallel, the ICD-11 officially recognized in 2018 gambling disorder and gaming disorder as addictions (World Health Organization, 2018). At the same time, the scientific literature accumulates behavioral, psychological and neurobiological data suggesting similarities between substance and behavioral addictions (Potenza, 2009; Kim and Hodgins, 2018).

In the media and in popular discourse, addiction also became a term, sometimes overused, to describe excessive engagement in everyday behaviors such as sugar consumption, screen use, or physical exercise. This overuse may reflect the fact that rewarding behaviors, much like the use of rewarding substances, can lead to repeated use, occasionally beyond what is considered reasonable, and sometimes difficult to control. This raises important questions: by which criteria can an addiction be identified? Is the phenomenon the same for substances and behaviors?

The inclusion of some behaviors in diagnostic classifications of addiction has extended the scientific interest toward other everyday rewarding activities that may become excessive and cause distress or impairment. This trend has also been amplified by the rise of the internet, which facilitates access to some of these behaviors (Young, 1998; Kuss and Griffiths, 2012). Consequently, a wide range of rewarding behaviors or activities, which can sometimes lead to excessive use, both online (e.g., compulsive buying, social media use, cyberchondria, or pornography use/compulsive sexual behavior), or offline (e.g., excessive exercise, work, or eating), have been proposed as potentially at risk for developing a behavioral addiction (Demetrovics and Griffiths, 2012; Brand et al., 2025). However, some authors warn against over-pathologization of common behaviors, and the risk of blurring the boundaries between “normal” and “pathological” behaviors, emphasizing that the frequency or amount of use is not sufficient to qualify for an addiction (Billieux et al., 2015). For example, excessive smartphone use or binge-watching remains debated as to whether they constitute true addictions (Panova and Carbonell, 2018; Ort et al., 2021). These examples highlight the urgent need to clarify the boundary between intensive, non-pathological engagement and a genuine pathology such as addiction.

To conceptualize and diagnose a potential behavioral addiction, several methods have been suggested. One method consists of exploring if the behavior fits into the construct of substance addiction, and transposes the SUD diagnostic criteria to this behavior (Grant et al., 2010). This “transdiagnostic approach” could be supported by overlapping symptoms and mechanisms observed between behavioral and substance addictions. An alternative approach is to examine this behavior as a new and distinct phenomenon, rather than as part of the addiction construct. This “behavior-specific approach” advocates the development of a behavior-specific diagnostic criteria set, grounded in a multidimensional understanding of the disorder, and validated through longitudinal data, psychometric analyses, and empirical research (Billieux et al., 2015, 2017; Castro-Calvo et al., 2021; Baggio et al., 2022). These two approaches have their advantages. The transdiagnostic approach could facilitate research and clinical recognition of behavioral addictions. This would provide a unified conceptual framework for all addictive behaviors, while building on established diagnostic structures and validated assessment tools. It would also facilitate the comparison of different addictions, accelerating the identification of common mechanisms and capturing the potential transdiagnostic core of addiction (Shaffer et al., 2004; Kim and Hodgins, 2018; Garrison et al., 2023). Conversely, the behavior-specific approach may help avoid over-pathologization by reducing the risk of false-positive diagnoses that could result from applying irrelevant criteria, while instead incorporating criteria that may be more specifically relevant to certain behavioral addictions (Billieux et al., 2015; Starcevic, 2016; Starcevic and Billieux, 2017; Castro-Calvo et al., 2021; Fineberg et al., 2022; Flayelle et al., 2025).

This debate over whether diagnostic criteria should be standardized across all addictions is far from new. Throughout the revision process related to the DSM-5, the transdiagnostic approach has been favored for the different substances, i.e., applying the same set of criteria to all SUD. This decision was grounded in the many common biological mechanisms, clinical symptoms and treatment responses across SUD, albeit the fact that not all individual criteria have the same psychometric characteristics when comparing between substances (Hasin et al., 2013; Koob and Volkow, 2016; Kervran et al., 2020; Volkow and Blanco, 2023).

We propose to reconcile these apparently divergent approaches (behavior-specific versus transdiagnostic) to conceptualize and diagnose behavioral addictions. We argue that such reconciliation may lie in a deeper understanding of the core impairment underlying addiction, whether substance-related or behavioral. Both the DSM-5 and ICD-11 emphasize that addictive use must be identified as problematic, i.e., causing significant distress and functional impairment, and rely on a concise set of central criteria regarded as the “core” features of addiction, primarily related to loss of control (Potenza, 2014; Koob and Volkow, 2016). In our opinion, the craving criterion may represent a promising avenue for capturing this loss of control from a transdiagnostic perspective.

Even though its definition remains debated, craving is generally described, in the field of addiction, as an intense desire or urge to use a substance (Sayette et al., 2000; Cherpitel et al., 2010; American Psychiatric Association, 2013), which persists beyond the withdrawal period, and that can be distinguished from a “normal” desire by its involuntary and unwanted nature (Auriacombe et al., 2018; Vafaie and Kober, 2022).

To avoid confusion, it is important to note that in English, craving is a term that is neither specific to addiction nor restricted to substance use. As such, craving is not pathological by nature, and could describe normal motivational and adaptive mechanisms that can emerge in response to any deprivation, such as food craving during fasting or craving for social contact during periods of isolation (Tomova et al., 2020). Adaptive and pathological craving are based on largely shared motivational circuits, and pathological craving in addiction is distinguished not by the nature of the mechanisms involved, but by its dysregulation, persistence beyond deprivation, and dominance over alternative rewards (Robinson and Berridge, 2008; Koob and Volkow, 2010; Volkow et al., 2016). The term craving is also used to describe compulsive urges (for food or particular eating behaviors) in eating disorders [for review (Cornil et al., 2026)], which interestingly have been frequently compared to addiction, and may partially overlap with food addiction (Meule and Kübler, 2012; Fauconnier et al., 2020; Mallorquí-Bagué et al., 2020; Chapron et al., 2023; Collombat et al., 2024). In addition, craving is also employed to describe anxiety-driven urges aimed at alleviating negative emotional states in obsessive–compulsive disorder (el-Guebaly et al., 2012; Jones and Bhattacharya, 2012). These forms of craving appear to differ in their underlying neurobiological substrates as well as in their responses to pharmacological treatment (el-Guebaly et al., 2012; Piquet-Pessoa and Fontenelle, 2016; Ferreira et al., 2021).

Accordingly, craving in its broad sense is not sufficient to distinguish addiction-related mechanisms from other motivational processes. Within the context of addiction, craving is more precisely conceptualized as a pathological motivational desire directed toward a particular rewarding object, characterized by its persistence independent of deprivation, its prioritization over alternative rewards, and its involuntary and unwanted nature (Robinson and Berridge, 2008; Koob and Volkow, 2010; Volkow et al., 2016; Auriacombe et al., 2018; Vafaie and Kober, 2022). In the remainder of this manuscript, the term craving refers to this definition of pathological craving in addiction.

At the diagnostic level, craving represents a core criterion across substance addictions. The introduction of craving as a diagnostic criterion for SUD in DSM-5, based on the growing body of evidence showing its major role in addiction, has sparked increased scientific interest (Sayette, 2016). Psychometric studies, such as Item Response Theory (IRT) analyses, examine how each criterion of a scale relates to (and is informative about) an underlying latent construct (Embretson and Reise, 2000; Sahin and Anil, 2017; Arend and Schäfer, 2019). IRT analyses have shown that craving was the least severe and most discriminant among the 11 DSM-5 SUD criteria when comparing across substances (Kervran et al., 2020) and samples of users (Chung et al., 2012; Gilder et al., 2014; Castaldelli-Maia et al., 2018; Kervran et al., 2021; Shmulewitz et al., 2023), suggesting that craving could represent an early manifestation of addictive processes, and could also be a sensitive marker of a transition from controlled to addictive use. More recently, some authors proposed that diagnostic criteria could be dynamically and causally mutually dependent, and thereby propose representation of a disorder as a network, i.e., a web of mutually influencing symptoms (Borsboom and Cramer, 2013), allowing exploration of their interrelationships (Fried, 2015; Epskamp et al., 2017; Schlechter et al., 2021). Such network analyses offer a window for formulating hypotheses about how disorders develop, are maintained and could be treated. Especially, centrality inferences can be understood to reflect how connected and thus potentially clinically relevant a diagnostic criterion is in a network. Such analyses applied to SUD criteria revealed that craving was a symptom that consistently remained in terms of centrality regardless of the substance explored, confirming its central place in the diagnosis of addiction (Gauld et al., 2023; Shmulewitz et al., 2025). These elements, observed across substances, support the hypothesis that craving is a transdiagnostic marker applicable to any substance addiction.

Interestingly, beyond its diagnostic value, craving is by nature a dynamic phenomenon whose frequency and intensity vary over time depending on internal and external factors (Sayette et al., 2000; Enkema et al., 2020). The use of Ecological Momentary Assessment (EMA), a method collecting information in natural contexts through repeated questionnaires completed directly on smartphone applications (Stone and Shiffman, 1994), has allowed to capture craving fluctuations in daily life (Serre et al., 2012, 2015, 2025). As a state, craving has shown prognostic relevance, as fluctuations in its intensity can predict subsequent substance use within a few hours, but also the risk of relapse over the course of several years (Fatseas et al., 2015; Serre et al., 2015, 2025; Auriacombe et al., 2018; Vafaie and Kober, 2022; Baillet et al., 2024a), documented for many different substance addictions, thereby confirming craving as a key target for intervention. However, craving does not always predict persistence of use, and its predictive value for relapse depends on both the timing and methodology of its assessment (Serre et al., 2015; Vafaie and Kober, 2022), as well as on its interaction with other psychological and contextual factors (Serre et al., 2025).

The craving phenomenon is also reported for behavioral addictions, including gambling disorder (Antons et al., 2023; Mallorqui-Bague et al., 2023; Westerberg et al., 2025), gaming disorder (Zhou et al., 2022; Park et al., 2023), problematic pornography use/compulsive sexual behavior (Antons et al., 2020; Antons and Matthias, 2020; Miele et al., 2023), food-related disorders (Roefs et al., 2019; Chapron et al., 2023), problematic social media use/problematic smartphone use (De-Sola et al., 2017; Moretta and Buodo, 2018) and excessive exercise (Heirene et al., 2016). Many authors highlight this phenomenon as relevant across behavioral addictions (Kim and Hodgins, 2018; Brand et al., 2019; Castro-Calvo et al., 2021). Similarly, despite the challenge of modeling craving in animals (Field and Kersbergen, 2020), some animal models of behavioral and non-substance addictions have reported craving-like responses to natural rewards such as sucrose or gaming (Grimm et al., 2005; Casile et al., 2025), suggesting that craving is neither intrinsically drug-specific nor necessarily dependent on direct external pharmacological effects or fundamental physiological needs. These models therefore offer the potential for an interesting framework to disentangle the components of craving that reflect adaptive motivational processes from those that index maladaptive, addiction-like mechanisms. However, despite being included in some assessment or screening tools for behavioral addictions (De-Sola et al., 2017; Panova and Carbonell, 2018; Fineberg et al., 2022; Chapron et al., 2023; Miele et al., 2023; Park et al., 2023; Antons et al., 2025), to date, craving has not been included as a diagnostic criterion for behavioral addictions in DSM-5/ICD-11 classifications.

At the neurobiological level, craving for different addictive behaviors has been associated with activation in rewards-related cerebral networks, particularly the ventral striatum and orbitofrontal regions involved in assigning value to stimuli. These motivational states also mobilize salience and emotional regulation circuits, such as the insula and anterior cingulate cortex, as well as prefrontal regions responsible for executive control (Potenza, 2008; Zhang et al., 2021; Park et al., 2023). Interestingly, these circuits show a large overlap with those involved in SUD.

At the behavioral level, craving for gaming, gambling, and social networks can be induced by cue exposure (Ashrafioun et al., 2012; Shin et al., 2018; Ciccarelli et al., 2019; Zhou et al., 2022), and craving intensity has been associated with severity and risk of relapse in gaming and gambling disorders (Quintero et al., 2020; Mallorqui-Bague et al., 2023; Westerberg et al., 2025). In daily life, EMA studies demonstrated that an increase in craving was associated with an increased probability of gambling (Dowling et al., 2021; Hawker et al., 2021) and pornography use (Henry et al., 2024) in the following hours, as it has been previously highlighted across substance addictions (Serre et al., 2015, 2018, 2025).

The findings presented above suggest that craving, well-characterized in SUD regarding neurobiological pathways and behavioral consequences, also constitutes an important feature of behavioral addictions. This leads us to propose that craving may represent a transdiagnostic cornerstone for a shared framework encompassing both substances and behavioral addictions, of value for both mechanistic understanding, diagnosis and treatment.

Several findings support the hypothesis that craving is a central characteristic across different forms of addiction. Psychometric analyses of a craving scale, applied to both substance and behavioral addictions, have identified craving as a latent factor shared across several addictions (Antons et al., 2025). A systematic review of craving assessment tools across substances and behaviors also supported the conceptualization of craving as a transdiagnostic phenomenon (Miele et al., 2023). At the neurobiological level, studies revealed the involvement of similar neural mechanisms and networks in predicting craving across different addictions (Antons et al., 2024). Converging evidence also emerged from deep transcranial magnetic stimulation studies, with reductions in craving obtained thanks to similar stimulation target patterns across both substance and behavioral addictions (Del Mauro et al., 2025). Nonetheless, further studies are needed.

At the diagnosis level, a first step could be to assess the diagnostic value of the craving criterion within the context of behavioral addictions, as it has been previously done to define the set of criteria for SUD in the DSM-5. For this purpose, a set of criteria, including both SUD-derived criteria and other potentially relevant ones specific to each addictive behavior (i.e., derived from qualitative interviews) could be assessed in a large sample of people who regularly engage in the behavior in question, for psychometric analyses. In the context of natural behaviors, because craving may overlap with adaptive motivation, its definition should be explicitly specified at the time of measurement. In SUD, IRT analyses have consistently demonstrated a unifactorial dimension and identified craving as one of the most discriminating and least difficult criteria to meet (Kervran et al., 2020). Similarly, when the DSM-5 criteria were adapted to food, unidimensionality was also confirmed and craving showed good discriminative value (Chapron et al., 2023). Drawing on these results, we hypothesize that the craving criterion will also appear to be one of the most psychometrically relevant criteria in behavioral addictions. In the same way, network analyses have highlighted the centrality of craving within the criteria networks of several SUDs (alcohol, cannabis, tobacco, opioids, and cocaine) (Gauld et al., 2023; Shmulewitz et al., 2025). Extending network analyses to behavioral addictions would allow to examine how craving is central within these disorders.

Beyond its value and utility for establishing diagnosis, recognizing the relevance of craving for behavioral addictions will also rely on demonstrating its etiological and predictive role in behavioral addictions. An important further step will be to explore how craving fluctuations relate to engagement in addictive behaviors and relapse. As discussed previously, several studies have begun to link craving intensity to addictive behavior practice, based on both laboratory and daily life data (Dowling et al., 2021; Hawker et al., 2021; Zhang et al., 2021; Ciccarelli et al., 2022; Zhou et al., 2022; Henry et al., 2024). The use of new approaches combining EMA and dynamic network analyses would make it possible to assess, at the within-person level, how craving interacts with other components (emotional, contextual, or cognitive components) that may influence behavior or relapse. In substance addiction, dynamic network analyses have shown that higher craving, combined with lower self-efficacy, are the best predictors of substance use in the following hours (Serre et al., 2025). We hypothesize that, in behavioral addictions, craving may retain this central role. Comparing substance and behavioral addictions networks could also help reveal factors that exert similar, or, conversely, distinct, influences, thereby contributing to a clearer identification of the central and common core that may support the development of a unified framework for all addictions.

Another important step to demonstrate the transdiagnostic value of craving across substance and behavioral addictions will be to explore if it shares biological substrates in both SUD and behavioral addictions. Some studies have already highlighted similarities in the neurobiology associated with craving (Antons et al., 2024; Del Mauro et al., 2025). In SUD, several studies have shown that craving was also associated to physiological changes in the autonomic nervous system (heart rate, electrodermal activity), suggesting the existence of a physiological signature specific to craving (Kennedy et al., 2015; Carreiro et al., 2020; Baillet et al., 2024b). These results support the idea of a biological marker for craving, measurable in daily life, but it remains necessary to extend these investigations to behavioral addictions to determine whether a comparable signature can be observed.

This paper has examined contemporary issues related to behavioral addictions, emphasizing the central and transdiagnostic role of craving.