RESEARCH DOSSIER
Proposed Behavioral Addictions and Diagnostic Validation
Abstract. Proposed behavioral addictions comprise patterns of harmful recurrent engagement in shopping, social media, smartphone-mediated activities, eating, and other domains whose status as independent addictive disorders is contested or incompletely established. This fundamental research monograph examines the difference between an informative psychological construct, a reliable screening measure, a validated clinical syndrome, and a recognized diagnosis. Through critical narrative synthesis of diagnostic standards, historical theory, psychometric validation, semistructured interviews, and longitudinal research through November 2026, it compares the evidence for gambling and gaming as classification anchors with the distinct evidential position of buying-shopping disorder, problematic social-media and smartphone use, and food-related addiction propositions. Particular attention is given to differential diagnosis, independent impairment measurement, item transfer, confounding, response biases, population transportability, and clinical utility. The overall synthesis supports careful investigation of serious behavioral dysregulation while requiring independent, prospective, and clinically meaningful validation before diagnostic categories are generalized.
Keywords: behavioral addiction; diagnostic validation; compulsive buying-shopping disorder; problematic social media use; smartphone overuse; food addiction; eating addiction; construct validity; differential diagnosis; functional impairment; DSM-5-TR; ICD-11; psychometrics; longitudinal research; clinical utility; Addiction Psychology
The Research Object and Diagnostic Question
Proposed behavioral addictions are recurrent patterns of engagement in activities for which researchers have suggested an addiction-like syndrome, while their status as distinct mental disorders remains disputed, incompletely validated, or dependent on the classification system. The principal problem is diagnostic validation: determining whether a proposed syndrome reliably identifies a clinically important condition, distinguishes affected people from high-engagement but healthy populations, predicts meaningful outcomes, and improves decisions about care. The research object therefore concerns the inference from behavior to disorder, not merely the intensity or popularity of any given activity.
Three phenomena must be considered together: the behavior itself, its functional organization over time, and the consequences attributed to it. Shopping can be financially ruinous, social-media engagement can displace sleep and responsibilities, and eating may be experienced as compulsive. These facts warrant investigation and assistance where harm occurs. Their interpretation requires a causal and diagnostic analysis: the same observed outcome can arise through mania, depression, anxiety, loneliness, occupational obligation, social pressure, restrictive eating, an exploitative digital environment, or a proposed addiction process. A valid diagnosis must do more than redescribe these possibilities in the vocabulary of addiction.
This monograph is a critical narrative scientific synthesis of diagnostic manuals, primary psychometric studies, longitudinal evidence, clinical interview research, and competing theoretical analyses. It neither implements a preregistered systematic-review search nor pools effect sizes in an original meta-analysis. Its conclusions are therefore conceptual and evidence-calibrated, with the strongest inferences reserved for designs that directly test their relevant claims. The review covers literature available through November 2026 and concentrates on compulsive buying-shopping, social-media and smartphone use, food-related propositions, and selected adjacent diagnostic controversies.
Historical Formation of Behavioral Addiction Research
Modern proposals developed partly through extensions of addiction theory beyond ingestion of substances. Shaffer and colleagues (2004) Shaffer, H. J., LaPlante, D. A., LaBrie, R. A., Kidman, R. C., Donato, A. N., & Stanton, M. V., 2004 formulated a syndrome model in which apparently different expressions of addiction may share etiological properties. Griffiths (2005) Griffiths, M., 2005 articulated a components model involving salience, mood modification, tolerance, withdrawal, conflict, and relapse. Both helped create a cross-domain research vocabulary, although a list of analogous symptoms is a hypothesis about shared structure rather than a definitive diagnostic test.
The subsequent decade brought widening applications of addiction language to internet use, video games, shopping, sexual behavior, work, exercise, social networking, and eating. This expansion generated a methodological debate about applying the same questionnaire items to any high-frequency activity. Billieux and colleagues (2015) Billieux, J., Schimmenti, A., Khazaal, Y., Maurage, P., & Heeren, A., 2015 argued that atheoretical confirmation and the routine interpretation of intensive habits as addictions could pathologize everyday life. Kardefelt-Winther and colleagues (2017) Kardefelt-Winther (2017) advanced an operational definition grounded in persistent functional harm and explicitly argued for exclusion criteria and alternative explanations.
The history of formal recognition developed more selectively. Gambling moved into the DSM-5 family of substance-related and addictive disorders. Internet gaming disorder was designated for further research. The World Health Organization subsequently adopted both gambling disorder and gaming disorder in ICD-11. This combination of expanding scholarship and selective nosological adoption is the historical foundation of the present problem. Classification decisions make important phenomena visible, yet they do not establish that every activity sharing surface features is clinically equivalent.
Current Diagnostic Classification: DSM-5-TR and ICD-11
The American Psychiatric Association's DSM-5-TR account American Psychiatric Association., n.d. identifies gambling disorder as its recognized behavioral addiction and locates internet gaming disorder among conditions recommended for further study. General smartphone or social-media addiction is not thereby an established DSM diagnosis. In the WHO description of ICD-11 gaming disorder World Health Organization (2020), the core elements are impaired control, increasing priority over other activities, and continuation or escalation despite negative consequences, with clinically significant functional impairment and a normally sustained pattern of at least 12 months.
These systems are related but do not use identical rule sets. DSM-5-TR internet gaming disorder retains a proposed symptom framework, whereas ICD-11 gaming disorder uses a narrower diagnostic formulation centered on control, priority, persistence, and harm. Researchers cannot transfer a questionnaire threshold from one system to the other without fresh validity evidence. The WHO Clinical Descriptions and Diagnostic Requirements (2024) World Health Organization., 2024 provide the internationally relevant clinical framework; national implementation and coding practice require further jurisdiction-specific attention.
Compulsive buying-shopping is an instructive borderline case. In ICD-11 it is discussed under the residual 'other specified impulse control disorder' category rather than established as a distinct named addiction with independently codified criteria, as described in 2026 classification research Müller (2026). Compulsive sexual behavior disorder is in ICD-11's impulse-control grouping. These classifications must not be rewritten as official decisions that every compulsive behavior is an addictive disorder. Terminological precision is essential when counseling clients, interpreting prevalence studies, or comparing international datasets.
The Meaning of Diagnostic Validation
Diagnostic validation concerns whether a set of proposed criteria corresponds to a meaningful clinical construct. It integrates content validity, internal structure, convergent and discriminant validity, criterion relationships, longitudinal prediction, clinical utility, and fairness across populations. A measure can show internal consistency and still fail to represent a distinct disorder. A model can fit a questionnaire dataset while embodying a misleading theory. Similarly, a group with higher average distress can be identified statistically without a defensible clinical threshold separating disorder from normal variation.
The distinction between validation and classification is crucial. A research construct may have established correlates, stable factor structure, and reliable self-report items long before sufficient evidence supports its placement in an international diagnostic manual. Classification additionally involves differential diagnosis, care consequences, reproducibility across clinical settings, and the balance between underidentification of suffering and overdiagnosis. The 2017 operational framework Kardefelt-Winther (2017) placed persistence, harm, and exclusionary reasoning at the center of this challenge.
A satisfactory validation program must specify exactly what phenomenon is being measured: engagement, loss of control, subjective distress, dependence on a particular platform, harmful financial exposure, or a distinct addiction-like syndrome. Studies should test these propositions against one another. Diagnostic circularity occurs when 'addiction' is defined by a questionnaire and that questionnaire is then used as the principal proof that the condition exists. Independent clinical interviews, time-ordered impairment, competing diagnostic measures, and response to targeted interventions provide stronger tests.
Disorder, High Engagement, and Functional Impairment
High engagement often reflects opportunity, necessity, identity, skill, community, or sustained interest. Students may rely on social networking for school, workers may require smartphones, and caregivers may spend long periods coordinating online. Engagement becomes relevant to diagnosis when contextualized by a loss of agency, recurring harm, persistent priority conflict, and failure to adjust behavior as circumstances change. Total hours are an exposure variable. They cannot by themselves identify a syndrome.
Functional impairment requires examination of important domains such as educational progress, occupational responsibility, financial stability, relationships, health, and daily living. It must be both significant and plausibly connected to the proposed behavior. This attribution is difficult in cross-sectional surveys: depression may increase online withdrawal while problematic online interactions may aggravate depression; financial stress can precede excessive shopping or follow it. A symptom checklist collected on one occasion cannot resolve temporal direction.
Clinical harm also varies in character. Financial debt from shopping, sleep deprivation from nighttime engagement, and nutritional or medical complications associated with eating differ in their pathways and available remedies. Collapsing them into a single global 'addiction severity' score may obscure clinical targets. The functional-impairment criterion should be measured independently of diagnostic symptoms whenever possible; otherwise the apparent association between symptoms and harm may be partly tautological.
Competing Explanations and Differential Diagnosis
A robust diagnosis withstands comparison with alternative explanations. For shopping, investigators must consider elevated mood or manic episodes, compulsive rituals, material insecurity, impulsivity, hoarding, and social or financial pressures. For social media, possible explanations include generalized anxiety, depressive avoidance, social reassurance seeking, attentional difficulties, school or occupational expectations, and exposure to harassment. For problematic eating, restrictive cycles, binge-eating disorder, trauma-related responses, metabolic states, and food insecurity can influence behavior without implying that a named substance-like addiction has been established.
Differential diagnosis is a scientific test of construct distinctness. A valid syndrome should identify clinically relevant variance beyond closely neighboring recognized disorders and transdiagnostic mechanisms. It should clarify whether the behavior is primary, a coping response, part of another syndrome, or a predictable reaction to an environment. The presence of comorbidity does not automatically invalidate a diagnosis, but a syndrome that cannot be distinguished from its proposed alternatives requires further theoretical specification.
Judgment must also account for personal values and social norms. Disapproval of a behavior can heighten self-reported guilt without establishing a disorder. Research on self-perceived pornography addiction, including Grubbs, Kraus, and Perry (2019) Grubbs, J. B., Kraus, S. W., & Perry, S. L., 2019, demonstrates the importance of studying moral incongruence together with behavior and impairment. Sexual behavior and pornography use require careful interpretation because ICD-11 compulsive sexual behavior disorder is categorized as an impulse-control disorder and the clinical assessment of distress must distinguish impairing dyscontrol from moral condemnation alone.
Components Models: Explanatory Reach and Diagnostic Risks
The components approach offers a shared vocabulary for evaluating recurrent activities. Salience captures prominence in thought and routine; mood modification describes anticipated emotional effects; tolerance concerns changing intensity or dose; withdrawal-like responses may include irritability when prevented from engaging; conflict denotes functional competition; and relapse marks return after efforts to reduce participation. Griffiths (2005) Griffiths, M., 2005 presented these as common components within a biopsychosocial framework.
Their meaning varies sharply across activities. Spending more time in an enjoyable hobby can reflect expertise rather than tolerance. Feeling disappointed when a communication device is unavailable can be a rational response to losing essential contact rather than clinical withdrawal. Relapse requires a prior problematic pattern and a meaningful goal of reduction; returning to a rewarding routine is not intrinsically pathological. These conceptual distinctions affect item wording, score interpretation, and false-positive rates.
The model remains valuable for developing testable comparisons. Investigators can ask whether a proposed behavior shows the same longitudinal sequence of craving, impaired control, harm, and recurrence as an established disorder, controlling for general distress and exposure. Convergent patterns support a process hypothesis. Demonstrating a new diagnostic category requires additional evidence for specificity, threshold validity, and clinical utility.
The I-PACE Model and Process-Level Explanation
The Interaction of Person–Affect–Cognition–Execution model, updated by Brand and colleagues (2019) Brand, M., Wegmann, E., Stark, R., Müller, A., Wölfling, K., Robbins, T. W., & Potenza, M. N., 2019, describes how predispositions, emotional responses, cognitive appraisals, cue-related craving, and executive processes can interact in the development and maintenance of selected addictive behaviors. Its process formulation is useful because diagnostic symptoms alone do not explain why a person returns to a harmful activity, how contextual cues acquire motivational force, or why the balance of deliberative and automatic control may vary.
The model suggests differentiated predictions. A person using social media primarily for reassurance may show stronger responses to interpersonal uncertainty, while a person purchasing goods to relieve negative affect may be especially sensitive to distress and the immediate reward of acquisition. A prospective study could test whether cue responses, affective shifts, and inhibitory-control measures predict later losses of control beyond baseline symptoms. Failure to support a proposed pathway would require revision of the mechanism rather than automatic rejection of the person's reported suffering.
A mechanistic model has to be evaluated independently of diagnostic labels. Similar neural or cognitive correlates across behaviors do not establish clinical equivalence; the same circuitry contributes to ordinary motivation and learning. A valid explanatory account should identify conditions under which processes predict significant harm and conditions under which equally intense engagement remains adaptive.
Compulsive Buying-Shopping: Conceptual History and Clinical Phenotype
Compulsive buying-shopping has a long descriptive history involving recurrent acquisition, difficulty resisting urges, affective relief, and adverse financial or interpersonal outcomes. Modern studies differentiate purchasing for utility, collecting, impulsive shopping, emotion regulation, and recurrent excessive buying despite harms. Online commerce adds rapid purchasing, individualized advertising, continuous availability, frictionless payment, and social comparison, each potentially affecting exposure and control. These environmental features should be measured as explanatory conditions rather than absorbed into an undifferentiated trait score.
Müller and colleagues (2021) Müller (2021) conducted a Delphi expert-consensus study that proposed diagnostic features including recurrent urges, diminished control, excessive purchase of unused items, mood regulation, impairment, cessation-related symptoms, and continuation despite consequences. This was an important operationalization step. Consensus increases clarity and enables comparable studies, but it does not itself establish population prevalence, causal mechanisms, or the status of a formally recognized independent diagnosis.
The clinical phenotype should be assessed through behavior-specific interviews about purchase patterns, unused inventory, debt, concealment, relationship conflict, failed limits, distress, and attempts at change. Interviewers should establish whether financial damage is attributable to buying behavior, debt structure, income instability, or multiple causes. Economic conditions and cultural norms alter what 'excessive' purchasing means; the validity of indicators must therefore be examined beyond affluent convenience samples.
A 2026 Interview-Based Test of Buying-Shopping Criteria
A major subsequent contribution is Ni, Jones, David, Müller, and Norberg (2026) Ni, N., Jones, S. B., David, J., Müller, A., & Norberg, M. M., 2026, who studied 51 participants reporting lived experience of compulsive buying-shopping disorder. Using semistructured interviews and directed content analysis, the researchers examined how the 2021 consensus criteria corresponded to participants' descriptions. Their abstract reports recurring buying, reduced control, mood-related purchasing, significant distress or impairment, and persistence despite consequences as common features. Preoccupation, tolerance, and withdrawal were less common and were proposed as supplementary rather than essential features.
This study moves beyond expert agreement toward direct phenomenological evidence. The small, self-selected clinical-experience sample, however, does not determine representative prevalence, sensitivity, specificity, or whether the criteria distinguish compulsive shopping from other conditions. Its methods principally address content relevance and experiential fit. A structured diagnostic interview, prospective field validation, and independent clinical comparator groups remain essential before wider diagnostic conclusions can be drawn.
The study also illustrates why criteria should not be imported mechanically from substance disorders. If tolerance and withdrawal-like experiences are variable, making them mandatory may exclude people with severe buying-related impairment; making common nonclinical experiences diagnostic may produce the opposite error. Empirically derived item weights and transparent clinical thresholds are necessary to resolve this measurement problem.
Buying-Shopping and the Limits of Impulse-Control Categories
The current ICD-11 treatment of compulsive buying-shopping within a residual impulse-control category highlights an unresolved nosological choice. The question is whether failure of inhibition constitutes the dominant mechanism, whether negative reinforcement and cue-dependent craving make an addiction model more informative, or whether shopping difficulties form a heterogeneous set of trajectories. The 2026 analysis of classification Müller (2026) explicitly situates this debate against ICD-11's residual category.
Each interpretation predicts different evidence. An impulse-control account emphasizes the relationship between urges, inhibition, and action. An addiction account emphasizes recurrent reinforcement, persistence despite harm, and conditioned motivational processes. A mood-regulation account suggests that changes in affect may precede purchasing and that intervention effects may operate through coping and environmental constraint. These mechanisms can coexist. Their relative contribution must be established in carefully characterized groups rather than inferred from terminological similarity.
The comparative evidence strengthened in 2026 when Müller and colleagues Müller (2026) studied differences among non-problematic, risky, and pathological buying-shopping groups. Their report describes 241 participants in the analyzed sample, including 62 classified with pathological buying-shopping, 62 with risky buying-shopping, and 117 controls; classification used standardized diagnostic interviews. Group differences in reported functional harm, shopping-related craving, and affective or cognitive features support a clinically meaningful pattern. The retrospective comparison and group-selection procedures limit causal inference, while the markedly unequal gender composition across groups constrains generalizability. The authors also acknowledge overlap between the diagnostic interview's group classification and subsequent harm measures, which requires caution about independent validation. This study is particularly important because it includes a risky buying-shopping comparator rather than relying only on extremes of healthy controls and severe cases.
Clinical utility may eventually favor a unified label, meaningful specifiers, or differentiated subtypes. Comparative diagnostic interviews, independent outcomes, and treatment-response moderation should guide that decision. Researchers should avoid presenting a proposed addiction formulation as though it already has the independent international classification accorded to gambling or ICD-11 gaming disorder.
The Social Media Disorder Scale and Psychometric Interpretation
Van den Eijnden, Lemmens, and Valkenburg (2016) van den Eijnden, R. J. J. M., Lemmens, J. S., & Valkenburg, P. M., 2016 developed a nine-item Social Media Disorder Scale based on the nine proposed DSM-5 criteria for internet gaming disorder. Their original study used three online surveys with 2,198 Dutch adolescents aged 10–17 and reported favorable structural validity, consistency, test–retest properties, and relationships with criterion measures. These findings support a usable research instrument in that setting; they do not independently establish an official social-media-use diagnosis.
A key methodological challenge follows from the instrument's genealogy. When items are transferred from gaming to social media, correlations among them can demonstrate that respondents endorse an internally coherent cluster. Whether the cluster measures the same latent construct as gaming disorder, a general dysregulation factor, adolescent distress, or platform-specific demands must be independently tested. Strong reliability is compatible with a construct having several plausible interpretations.
Cross-validation should include independently rated impairment, clinical interviews, repeated measures, alternative disorders, and tests of differential item functioning by age, language, culture, and platform access. Thresholds optimized in one adolescent sample should not be treated as universal clinical cutoffs. Changing platform practices also create a longitudinal measurement problem: a symptom item about 'preoccupation' may have different normative meaning in school communities where online contact is required.
Longitudinal Findings in Social-Media Research Through 2026
Longitudinal research addresses temporal relationships that cross-sectional association cannot resolve. Remondi and colleagues (2026) Remondi (2026) examined 1,562 Italian adolescents over three waves across one year using a random-intercept cross-lagged panel model. The reported associations included stable between-person correlations and bidirectional within-person links between changes in problematic social-media use and emotional symptoms. The study therefore supports a dynamic relationship while leaving identification of a distinct addictive disorder an additional step.
The longitudinal psychometric study by Spyropoulou and Mastrotheodoros (2026) Spyropoulou, E., & Mastrotheodoros, S., 2026 followed 306 Greek early adolescents across three waves. It reported longitudinal structural stability alongside variation in item information, with greatest measurement precision around moderate problematic-use levels. The study supports the usefulness of the instrument for its evaluated research context and warns against assuming equal discrimination throughout the extreme clinical range.
Longitudinal correlations remain observational. Reverse influence, unmeasured time-varying confounding, and changes in platform features can complicate causal interpretation. Stronger diagnostic evidence would connect independently assessed symptoms to subsequent functioning and demonstrate incremental prediction beyond baseline mood, prior functioning, and social context. A high score that precedes depression may indicate risk without specifying the causal pathway or an independently classifiable addiction.
Smartphone Overuse and the Problem of the Device as Target
A smartphone is a multifunctional access device. It mediates education, work, transportation, financial management, family contact, gaming, social media, photography, and emergency communication. A person who spends substantial time with a smartphone may be engaged in one persistent activity or many unrelated tasks. Treating the device itself as a unitary addictive object is therefore a substantive claim requiring unusually careful behavioral disaggregation.
Panova and Carbonell (2018) Panova, T., & Carbonell, X., 2018 reviewed the smartphone-addiction literature and argued that the evidence available did not justify confirming smartphone addiction as a separate disorder. They recommended investigating problematic use and the sociocultural context of compensatory functions and motivations. Their conclusion should be understood as a critical scientific position, not as evidence that problematic technology use can never lead to serious impairment.
Measurement should distinguish applications and purposes, passive from interactive use, occupationally mandated engagement, nighttime interruption, and stress-related checking. Device telemetry can improve exposure measurement yet still cannot reveal the meaning of an action or the existence of clinically significant loss of control. A credible study combines behavioral logs with interviews, contextual diaries, and functional outcomes, while explicitly investigating how device restrictions might impair legitimate responsibilities.
Food Addiction, Eating Addiction, and Competing Diagnoses
The proposed food-addiction construct attempts to explain a subset of recurrent eating behaviors through addiction-related mechanisms. Its scientific evaluation faces a distinctive challenge: eating is biologically necessary, and food reward is intertwined with hunger, metabolic regulation, stress, restriction, affordability, availability, and culture. Establishing a distinct addictive syndrome requires discriminating problematic patterns from binge-eating disorder, bulimia nervosa, ordinary preferences, food insecurity, and metabolic or medication-related changes.
Gearhardt, Corbin, and Brownell (2016) Gearhardt, A. N., Corbin, W. R., & Brownell, K. D., 2016 developed the Yale Food Addiction Scale Version 2.0 as a measure of addictive-like eating behavior adapting criteria from substance-use disorders. Validation of such a scale supports research on the measured symptom pattern. It does not establish that a particular food, ingredient, or class of foods has been shown to function in humans as a substance of abuse in the pharmacological sense. The causal role of processing, palatability, scarcity, restriction, and eating routines requires separate investigation.
Hebebrand and colleagues (2014) Hebebrand (2014) argued that 'eating addiction' may be more conceptually appropriate than 'food addiction' because evidence for addiction to specific macronutrients in humans was insufficient. These competing interpretations point to different research objects. Clinical studies must specify whether they investigate compulsive behavioral organization, a response to a food environment, an existing eating disorder, or exposure-related reinforcement. Diagnostic validation is therefore inseparable from careful characterization of the target.
Further Diagnostic and Methodological Analysis
The preceding case studies reveal a shared validation problem. Proposed syndromes are easiest to identify at the level of surface experience: persistent urges, attempted limits, and troubling consequences. The scientific challenge is to determine which combinations of these elements form a stable clinical condition, how they arise, which alternative explanations they exclude, and whether naming a disorder improves practical decisions. The next sections examine research designs capable of answering those questions.
A coherent research strategy should proceed from clearly specified behavior and independently verified impairment through reliable assessment, clinical differentiation, and time-ordered outcomes. Models of motivation, cue learning, executive control, and environmental reinforcement can guide hypotheses, while classification decisions should be revised as disconfirming evidence accumulates. The theoretical strength of the field depends on accepting the possibility that some proposed disorders will be validated, others will remain dimensional problems, and some will be better understood as expressions of other conditions.
Construct Validity, Reliability, and Diagnostic Accuracy
Measurement reliability answers whether observations are reproducible under appropriate conditions. Construct validity asks what those observations mean. Diagnostic accuracy asks how well a specified threshold identifies people who meet a credible clinical reference standard. These are separate questions. A questionnaire with excellent internal consistency can overcount normal use; a stable factor structure may reflect shared wording; good correlation with anxiety or depression can support convergent relevance while raising concerns about discriminant validity.
A complete measurement evaluation should estimate dimensionality, test–retest stability, measurement invariance, sensitivity, specificity, positive predictive value, negative predictive value, and classification uncertainty. Interpretive emphasis varies by purpose. A broad screening instrument may tolerate some false positives provided it triggers a careful follow-up assessment. A diagnostic label with lasting treatment, insurance, educational, or legal consequences demands higher specificity and transparent uncertainty. The positive predictive value of any cutoff depends strongly on disorder prevalence, which is often poorly known for proposed syndromes.
Where no credible reference standard exists, validating a score against a clinician judgment that itself relies on the same score creates incorporation bias. Blinded independent assessment, alternate information sources, and prospective functional outcomes are essential. Researchers should report how thresholds were selected, whether they were tuned in the same sample used to evaluate them, and how results change under plausible alternative cutoffs. Diagnostic validation cannot be replaced by reporting a single coefficient or favorable model-fit statistic.
Item-Response Theory, Threshold Selection, and False Positives
Item-response theory allows researchers to ask where along a latent severity dimension each symptom provides information. Items may be informative for moderate problems yet insensitive at the severe tail, precisely where clinical classification would be most consequential. The study by Spyropoulou and Mastrotheodoros (2026) Spyropoulou, E., & Mastrotheodoros, S., 2026 of the Social Media Disorder Scale is relevant because its reported precision was highest at moderate levels. A reliable total score is therefore not equivalent to a validated distinction between high engagement and serious disorder.
A threshold should be evaluated against independent severity and functioning, not chosen merely because a selected percentage of respondents endorse enough items. Receiver-operating-characteristic analysis requires a defensible reference classification. Researchers should report uncertainty intervals, subgroup performance, calibration, and external validation. Prevalence estimated from a provisional cutoff is conditional on that cutoff; comparisons across studies using incompatible thresholds are particularly vulnerable to false precision.
Item content also requires clinical scrutiny. 'Thinking frequently about an activity' is common among enthusiasts; 'remaining online longer than intended' may reflect a platform or work requirement; 'becoming irritable when disconnected' may indicate frustration, social dependency, or fear of missing necessary communication. Item-response models improve statistical calibration only after their semantic and causal interpretation has been examined.
Sampling, Selection Bias, and Population Transportability
Convenience online surveys are useful for theory development but frequently overrepresent highly engaged users and people motivated to report problems. Clinical cohorts may capture more severe experiences while excluding individuals who experience substantial harm yet never seek care. An interview study recruiting people who already identify with a named disorder can test whether proposed criteria describe their experiences; it cannot determine how common that disorder is in the general population. The contrast between the 2021 buying-shopping Delphi study Müller (2021) and the 2026 interview study Ni, N., Jones, S. B., David, J., Müller, A., & Norberg, M. M., 2026 illustrates distinct and complementary forms of evidence.
Representative probability sampling, stratification by age and socioeconomic position, and transparent attrition analyses improve population inference. Instruments also need translation and cultural adaptation that preserve the meaning of symptoms. A high score for 'neglecting offline social activities' may carry different significance where work, education, or community participation occurs primarily online. Financial harm from shopping requires evaluation relative to resources and obligations.
Transportability is a clinical question as well as a statistical one. Researchers should assess performance across adolescents and adults, among people with disabilities, in different health-care systems, and across changing digital markets. Replication in another sample from the same recruitment platform offers weaker transport evidence than genuinely independent evaluation in a distinct context.
Longitudinal Validation and the Direction of Causation
Prospective designs offer the opportunity to separate stable vulnerability from within-person change. Repeated measures can establish whether reported loss of control precedes impairment, follows emotional distress, or participates in a reciprocal process. Remondi and colleagues (2026) Remondi (2026) exemplify a stronger longitudinal strategy by using random-intercept cross-lagged modeling rather than interpreting cross-sectional correlations as temporal effects. Their reported bidirectional findings support investigating feedback processes without identifying one simple causal direction.
Time-varying exposures require careful modeling. Examinations, unemployment, relationship dissolution, major platform redesigns, changes in disposable income, and medication effects can influence both problematic behavior and functioning. If such factors are unmeasured, apparent prospective prediction remains open to confounding. Attrition may be systematically related to the very severe symptoms researchers wish to study. Preregistration and planned sensitivity analyses reduce selective interpretation.
A strong validation program would compare trajectories of high-engagement healthy participants, people with verified functional harm, and clinically characterized controls; estimate onset, persistence, remission, and recurrence; and test whether proposed criteria forecast outcomes beyond existing diagnoses. The clinically meaningful endpoint is not simply a high score at the next wave but sustained change in functioning and need for assistance.
Ecological Momentary Assessment and Behavioral Trace Data
Ecological momentary assessment captures behavior, emotion, and context closer to the time at which they occur. For shopping, it may record purchasing episodes, urges, affect before and after purchases, spending relative to available funds, and later regret. For digital engagement, it can capture notifications, purposes of use, self-imposed limits, and interference with sleep or responsibilities. This temporal resolution is valuable for distinguishing mood regulation, cue-triggered action, deliberate recreation, and habitual checking.
Digital trace data may estimate frequency, duration, application switching, and exposure to platform prompts. Such records are not a direct measure of craving, impaired control, or distress. A device cannot determine whether a prolonged online session is mandatory work, social support, avoidance, or compulsive engagement. Combining telemetry with interviews and functioning measures can reduce recall error while maintaining interpretive specificity.
Research ethics are inseparable from validity. Continuous monitoring may alter the behavior studied, create unequal burdens, or expose private communications. Consent, data minimization, limited retention, security, transparency about inferred attributes, and separation from punitive monitoring practices are substantive methodological requirements. Studies should report what behavioral logs can actually measure rather than interpreting digital intensity as a diagnosis.
Experimental Paradigms and Mechanism Testing
Laboratory paradigms can examine cue reactivity, reward expectancy, attentional capture, delay discounting, inhibition, and response to environmental friction. Such mechanisms become clinically relevant when they predict real-world impairment or change under targeted interventions. A cue-response difference between groups with high and low questionnaire scores may reflect familiarity, interest, or general emotional arousal. Evidence for addiction-specific pathology requires appropriately matched controls and functional endpoints.
Researchers can experimentally alter notifications, purchase friction, spending limits, content recommendations, or availability of alternative activities and assess their effects on both behavior and well-being. This approach tests environmental causation more directly than inferring neurobiological disorder from a correlational imaging contrast. However, a reduction in use after an intervention is not by itself proof that pre-intervention behavior was pathological.
Neurocognitive findings should be interpreted with special caution. Reward processing, salience detection, and inhibitory systems are involved in ordinary learning and goal pursuit. Similar neural responses across gaming, shopping, and substance disorders can support investigation of shared processes without constituting diagnostic biomarkers. Proposed disorders currently require careful clinical assessment; laboratory tasks do not substitute for that assessment.
Discriminant Validity and Transdiagnostic Processes
An independent diagnosis should provide clinically useful information that is not already captured by neighboring conditions. Studies can compare problematic buying with bipolar-spectrum symptoms, obsessive-compulsive phenomena, depression, material stress, and hoarding-related impairment. Social-media difficulties should be tested alongside anxiety, depression, attentional problems, loneliness, and social adversity. Eating-related constructs require particularly rigorous comparisons with recognized eating disorders and nutritional or metabolic factors.
Statistical adjustment for another questionnaire does not fully solve differential diagnosis. Multiple symptom scales can share response style, distress wording, and measurement error. A more persuasive approach combines structured interviews, multi-informant functioning, longitudinal course, and comparisons of the way individuals respond to specific interventions. A syndrome may still be useful when comorbid, but its proposed boundaries and added value must be demonstrated.
Broad vulnerability dimensions such as negative emotionality, impulsive decision-making, compulsive perseveration, and reduced coping flexibility may explain overlapping behavioral symptoms. Testing shared and distinct mechanisms can prevent unnecessary multiplication of diagnoses while preserving specific treatment targets for people whose shopping, online activity, or eating is genuinely impairing.
Clinical Significance and Patient-Centered Outcomes
Clinical significance belongs to the person's functioning and care needs. Diagnostic studies should ask whether the behavior interferes with relationships, economic security, health, autonomy, and meaningful activities over time. The severity of an experience cannot always be inferred from its frequency. Rare but catastrophic buying episodes may carry different risks from constant low-stakes browsing; social-media engagement may be extensive and protective for a geographically isolated individual.
Patient reports are indispensable but require interpretive context. Self-identification as 'addicted' may express a genuine sense of dyscontrol, a desire for help, a culturally available explanation, or moral discomfort. Grubbs, Kraus, and Perry (2019) Grubbs, J. B., Kraus, S. W., & Perry, S. L., 2019 demonstrate why self-perceived pornography addiction should be studied alongside actual use and values-related distress. A person should receive an appropriate assessment and support for severe impairment even while the most fitting diagnosis remains under scientific study.
Treatment response can contribute to validation when studies compare interventions against credible alternatives and measure changes in functioning. Improvement following cognitive behavioral therapy, motivational intervention, financial safeguards, or a change in digital environment does not uniquely validate an addiction framework: many interventions act on shared processes such as affect regulation and stimulus control. Clinical utility requires identifying which formulation best guides appropriate, effective, and proportionate care.
Prevalence, Base Rates, and Global Comparisons
Reported prevalence depends on the definition of cases, measurement instrument, age and access of the sample, threshold, reference period, and method of recruitment. A percentage of adolescents screening positive on a social-media questionnaire does not equal the prevalence of a recognized mental disorder. Likewise, the number of people who report frequent shopping urges cannot be treated as the number with a validated buying-shopping diagnosis.
False positives become particularly important when the underlying disorder is uncommon. Even a seemingly favorable test can have modest positive predictive value at low prevalence. A global survey using a uniform digital questionnaire may establish comparable patterns of responses while failing to secure equivalent clinical meaning across cultures and platforms. Survey reports should specify the exact measured construct instead of presenting an unqualified estimate of addiction.
Epidemiological maturity requires population-based validation with transparent weighting, response rates, clinically anchored thresholds, impairment measures, and uncertainty intervals. Researchers must also allow for changing availability and norms. A new application, economic shock, or legal restriction can change exposure and harms even when underlying vulnerabilities remain stable. Prevalence estimates are therefore dated observations of specified criteria in specified populations.
Comorbidity, Clinical Risk, and Safety
Proposed behavioral addictions intersect with emotional disorders, suicidality, financial distress, family conflict, sleep problems, and other clinical difficulties. Where serious impairment exists, the immediate responsibility is careful assessment of risks and needs, independent of whether a distinct addiction label has achieved diagnostic consensus. Compulsive spending may involve debt or exposure to unsafe borrowing; disrupted sleep may increase vulnerability; problematic eating may coexist with medical complications requiring established evaluation pathways.
Clinical claims should be linked to the particular population and design from which they arise. An adolescent association between social-media symptoms and depressive symptoms does not establish that any individual person's symptoms were caused by a platform. The 2026 Italian longitudinal study Remondi (2026) supports attention to reciprocal change while remaining observational. Similarly, associations between eating-related symptom scales and body weight cannot establish an addiction to a specific nutrient.
Assessment and care are best grounded in recognized comorbid conditions, demonstrable functional problems, patient circumstances, and current professional guidance. The present scholarly analysis offers no individual diagnosis, medication protocol, or unvalidated treatment promise. Clinical uncertainty should be represented accurately in records, referral pathways, and public communication.
Digital Environments, Platform Design, and Behavioral Opportunity
A platform shapes available choices through notification timing, reward schedules, personalized feeds, recommendation algorithms, purchase prompts, and social feedback systems. E-commerce and social media operate in commercial environments that can amplify exposure and reduce friction. These factors are relevant to the development of persistent behavior and may contribute to harm without establishing that every user exposed to them develops a disorder.
Behavioral opportunity provides an alternative analytic level to person-only explanations. An intervention that removes predatory design or improves purchase transparency may reduce harmful activity across people with different psychological vulnerabilities. King and Delfabbro (2018) King, D. L., & Delfabbro, P. H., 2018 discussed monetization mechanisms in games that blur boundaries between gaming and gambling. The broader lesson is that environmental affordances should be investigated alongside individual traits and clinical symptoms.
Digital platform changes also threaten the stability of diagnostic instruments. An item written for a previous media environment may change meaning when messaging, schooling, shopping, and entertainment converge in a single interface. Longitudinal comparability requires researchers to document changes in technology and conduct measurement-invariance checks. The scientific unit of analysis should be explicit: person, activity, platform, transaction, social interaction, or their reciprocal relation.
Youth, Families, and Developmental Context
Adolescents experience rapid changes in peer affiliation, self-regulation, autonomy, school demands, and access to digital environments. An instrument developed in young populations may capture important difficulties while also reflecting norms specific to developmental stage. The original Social Media Disorder Scale study van den Eijnden, R. J. J. M., Lemmens, J. S., & Valkenburg, P. M., 2016 evaluated Dutch adolescents. Subsequent longitudinal studies of Italian and Greek young people extend the empirical base but do not establish immediate equivalence to older adults or children of different ages.
Assessment should consider opportunities for offline connection, caregiver mediation, economic resources, educational expectations, disability accommodations, and the role of online communities. A young person's dependence on social connectivity may reflect isolation or limited alternative opportunities. Research therefore benefits from distinguishing self-reported distress, independently observed impairment, and contextual restrictions imposed by adults or institutions.
An ethically defensible developmental program evaluates both harms and benefits and follows trajectories into later life. It should compare adolescents with similar levels of digital exposure but differing functional outcomes, and should evaluate whether symptom thresholds predict enduring need for clinical assistance. Protective services and family supports should be offered according to demonstrable needs rather than a speculative label alone.
Cultural Validity, Norms, and Equity
Diagnostic categories travel across languages, socioeconomic systems, and normative environments. Their component items may not travel equivalently. Shopping-related financial consequences depend on income, debt systems, and family obligations. Social-media withdrawal may disrupt social participation in a community where communication is predominantly digital. Concerns about food-related control can intersect with cultural dietary rules and economic access to adequate nutrition.
Cross-cultural validation requires more than translating words. Interviews should establish the local meaning of symptoms; psychometric analyses should evaluate measurement invariance and differential item functioning; clinical studies should compare functioning with culturally appropriate reference standards. A threshold derived in a wealthy urban sample may misclassify patterns in rural, disadvantaged, or differently connected populations.
Equity also concerns the consequences of diagnosis. Labels can facilitate support, but an unsupported classification may amplify stigma, school discipline, insurance barriers, or unjustified surveillance. Conversely, dismissing all reports of dyscontrol because a condition is not formally recognized can leave severe functional needs unaddressed. The scientific goal is reliable, fair, and proportionate assessment based on observable impairment and evidence.
Open Science, Replication, and Publication Bias
The field is vulnerable to selective publication of statistically significant associations between new behaviors and familiar symptom domains. High correlations among overlapping self-report items, convenience recruitment, flexible threshold selection, and multiple exploratory comparisons can make a construct appear more settled than it is. The availability of large online samples increases statistical power without resolving questions of clinical meaning.
Preregistration should distinguish confirmatory diagnostic predictions from exploratory scale development. Replication should use independent samples, transparent analytic code, common outcome definitions, and direct tests of competing explanations. Negative findings deserve publication, particularly when a proposed syndrome fails to predict independent impairment or adds little to established diagnostic measures. Researchers should share instruments and scoring algorithms where licensing and privacy permit, and document deviations from planned analyses.
Evidence syntheses should report whether source studies involve diagnosed clinical participants, screening-positive community samples, qualitative interviews, or experimental paradigms. Pooling these categories without distinction can yield misleading prevalence or outcome statements. This monograph applies a critical narrative method and does not claim the completeness, quantitative aggregation, or formal risk-of-bias scoring of a systematic review.
Criteria for Future Diagnostic Recognition
Future diagnostic recognition should rest on convergent evidence from multiple domains. The proposed syndrome needs a clearly defined behavior and stable symptom structure; reliable clinical assessment; substantial and independently demonstrated impairment; persistent patterns over a meaningful period; differential separation from plausible alternatives; valid thresholds; prospective prediction; independent replication; population transportability; and demonstrable utility for clinical decisions.
These requirements should be treated as an empirical program, not as a box-ticking exercise. Strong evidence of impairment paired with weak evidence of construct distinctness may justify describing a problematic behavior and addressing its harms without a separate addiction label. Strong dimensional psychometrics without a defensible clinical threshold may justify a research scale and screening follow-up rather than categorical diagnosis. Evidence that a proposed syndrome changes prognosis and treatment matching beyond recognized conditions strengthens its case for classification.
Regulatory and classification bodies can then consider benefits and costs of recognition, including access to care, stigma, overdiagnosis, cultural effects, and public-health implications. The WHO's 2024 clinical requirements World Health Organization., 2024 and the contrasting history of gaming disorder illustrate that classification is both scientific and practical. Scientific validation remains an ongoing responsibility after inclusion in any manual.
Comparative Evidence Assessment
The evaluated research domains occupy different evidential positions. Gambling disorder provides the formal diagnostic anchor in DSM-5-TR and ICD-11. Gaming disorder is recognized in ICD-11, while internet gaming disorder remains a condition for further research in DSM-5-TR. Buying-shopping disorder has well-developed clinical descriptions and new interview-based validation but disputed classification and the need for structured field trials. Social-media research has meaningful psychometric and longitudinal advances, yet the independent diagnosis remains proposed. Smartphone overuse is still particularly challenged by the heterogeneity of activities accessed through the device. Food-addiction research has validated symptom measures while confronting unresolved questions of behavioral versus substance-based conceptualization and overlap with established eating disorders.
Evidence should be evaluated domain by domain, not granted through analogy. For buying-shopping, observational reports of harm and emerging consensus criteria merit targeted independent clinical studies. For social-media use, longitudinal within-person results justify further causal work and external impairment validation. For smartphone use, finer activity-level assessment is required. For food-related constructs, the proposed causal agent and differential diagnosis must be specified before strong nosological conclusions are appropriate.
Across all areas, the weakest inference is that endorsement of addiction-like language proves an addiction diagnosis. The strongest prospective path combines patient experience, operational clarity, independent measures of functioning, carefully chosen comparisons, and replication across populations. This synthesis favors diagnostic precision while preserving access to care for people experiencing significant harm.
Research Priorities for the Artificial Era
The Artificial Era adds new explanatory variables to the study of behavioral dysregulation: adaptive interfaces, generative companions, autonomous recommendation systems, interactive purchases, conversational persuasion, and personalized feedback. These systems may change exposure, social relationships, and the dynamics of cue engagement. Their presence calls for better measurement of interaction design, user goals, contextual dependence, and functional consequences.
For psychological classification, the relevant research question concerns human functioning in environments increasingly organized by artificial systems. Extended interaction with an artificial agent may support learning or emotional connection in one context and displace necessary activities in another. Time spent in an interface cannot establish clinical disorder without a validated syndrome, temporal evidence, and measured impairment. Anthropomorphic design and claims about artificial agency should be operationalized as variables rather than assumed explanations for every adverse outcome.
The historical horizon From Homo to Artificial provides a context for identifying new research settings. Empirical psychological conclusions remain grounded in testable claims about behavior, experience, environment, and outcome. Future studies can use transparent human–artificial interaction logs, consent-based prospective designs, experimental interface comparisons, and clinically meaningful measures to test specific pathways of harm and adaptation.
Conclusion
Proposed behavioral addictions and their diagnostic validation constitute a central methodological problem of contemporary psychology. The same behavioral pattern can express intense engagement, persistent dysregulation, adaptation to difficult circumstances, the consequences of commercial design, or another psychological disorder. The scientific task is to identify which configurations of symptoms, functioning, time course, and explanatory processes justify a distinct clinical category.
The strongest existing classification anchors remain gambling disorder and ICD-11 gaming disorder. Compulsive buying-shopping has developed a progressively clearer symptom description and new interview-based evidence, yet requires further independent validation. Social-media research increasingly demonstrates psychometric coherence and prospective associations with functioning and emotional symptoms, while the proposed diagnostic identity remains unsettled. Smartphone-centered labels face the conceptual problem of a device performing many functions. Food-related propositions remain scientifically productive but demand a sharper distinction among addictive-like eating, exposure to particular foods, and established eating disorders.
Diagnostic legitimacy emerges from evidence of reliable assessment, independent and clinically meaningful impairment, discriminant and predictive validity, causal understanding, replication, cultural transportability, and practical benefit. The field advances when researchers keep proposed constructs open to falsification and revise them in response to evidence. People whose behavior causes serious harm require thoughtful assessment and access to appropriate care throughout that scientific process.
References
- American Psychiatric Association. (2022). Diagnostic and statistical manual of mental disorders (5th ed., text rev.). American Psychiatric Association Publishing. https://doi.org/10.1176/appi.books.9780890425787
- American Psychiatric Association. (n.d.). Internet gaming. https://www.psychiatry.org/patients-families/internet-gaming/
- Billieux, J., Schimmenti, A., Khazaal, Y., Maurage, P., & Heeren, A. (2015). Are we overpathologizing everyday life? A tenable blueprint for behavioral addiction research. Journal of Behavioral Addictions, 4(3), 119–123. https://doi.org/10.1556/2006.4.2015.009
- Brand, M., Wegmann, E., Stark, R., Müller, A., Wölfling, K., Robbins, T. W., & Potenza, M. N. (2019). The Interaction of Person-Affect-Cognition-Execution (I-PACE) model for addictive behaviors: Update, generalization to addictive behaviors beyond internet-use disorders, and specification of the process character of addictive behaviors. Neuroscience & Biobehavioral Reviews, 104, 1–10. https://doi.org/10.1016/j.neubiorev.2019.06.032
- Gearhardt, A. N., Corbin, W. R., & Brownell, K. D. (2016). Development of the Yale Food Addiction Scale Version 2.0. Psychology of Addictive Behaviors, 30(1), 113–121. https://doi.org/10.1037/adb0000136
- Griffiths, M. (2005). A 'components' model of addiction within a biopsychosocial framework. Journal of Substance Use, 10(4), 191–197. https://doi.org/10.1080/14659890500114359
- Grubbs, J. B., Kraus, S. W., & Perry, S. L. (2019). Self-reported addiction to pornography in a nationally representative sample: The roles of use habits, religiousness, and moral incongruence. Journal of Behavioral Addictions, 8(1), 88–93. https://doi.org/10.1556/2006.7.2018.134
- Hebebrand, J., Albayrak, Ö., Adan, R., Antel, J., Dieguez, C., de Jong, J., Leng, G., Menzies, J., Mercer, J. G., Murphy, M., van der Plasse, G., & Dickson, S. L. (2014). 'Eating addiction,' rather than 'food addiction,' better captures addictive-like eating behavior. Neuroscience & Biobehavioral Reviews, 47, 295–306. https://doi.org/10.1016/j.neubiorev.2014.08.016
- Kardefelt-Winther, D., Heeren, A., Schimmenti, A., van Rooij, A., Maurage, P., Carras, M., Edman, J., Blaszczynski, A., Khazaal, Y., & Billieux, J. (2017). How can we conceptualize behavioural addiction without pathologizing common behaviours? Addiction, 112(10), 1709–1715. https://doi.org/10.1111/add.13763
- King, D. L., & Delfabbro, P. H. (2018). Predatory monetization schemes in video games (e.g. 'loot boxes') and internet gaming disorder. Addiction, 113(11), 1967–1969. https://doi.org/10.1111/add.14286
- Moretta, T., & Wegmann, E. (2025). Toward the classification of social media use disorder: Clinical characterization and proposed diagnostic criteria. Addictive Behaviors Reports, 21, 100603. https://doi.org/10.1016/j.abrep.2025.100603
- Müller, A., Brand, M., Thomas, T. A., Kessling, A., Schmid, A. M., Jiménez-Murcia, S., Wegmann, E., Müller, S. M., & Steins-Loeber, S. (2026). Empirical support for recognizing pathological buying/shopping as a mental disorder. Comprehensive Psychiatry, 146, 152673. https://doi.org/10.1016/j.comppsych.2026.152673
- Müller, A., Laskowski, N. M., Trotzke, P., Ali, K., Fassnacht, D. B., de Zwaan, M., Brand, M., Häder, M., & Kyrios, M. (2021). Proposed diagnostic criteria for compulsive buying-shopping disorder: A Delphi expert consensus study. Journal of Behavioral Addictions, 10(2), 208–222. https://doi.org/10.1556/2006.2021.00013
- Ni, N., Jones, S. B., David, J., Müller, A., & Norberg, M. M. (2026). Evaluating diagnostic criteria for compulsive buying-shopping disorder. Psychology of Addictive Behaviors. Advance online publication. https://doi.org/10.1037/adb0001166
- Panova, T., & Carbonell, X. (2018). Is smartphone addiction really an addiction? Journal of Behavioral Addictions, 7(2), 252–259. https://doi.org/10.1556/2006.7.2018.49
- Remondi, C., La Tona, A., Salerno, L., Aiello, R., Donisi, V., Poli, S., De Caro, E. F., Delvecchio, E., & Brugnera, A. (2026). Bidirectional longitudinal associations between problematic social media use and emotional symptoms in adolescents: Results from a random intercept cross-lagged panel model. Journal of Adolescence, 98(7), 2558–2566. https://doi.org/10.1002/jad.70237
- Shaffer, H. J., LaPlante, D. A., LaBrie, R. A., Kidman, R. C., Donato, A. N., & Stanton, M. V. (2004). Toward a syndrome model of addiction: Multiple expressions, common etiology. Harvard Review of Psychiatry, 12(6), 367–374. https://doi.org/10.1080/10673220490905705
- Spyropoulou, E., & Mastrotheodoros, S. (2026). Evaluating the longitudinal measurement properties of the Social Media Disorder Scale in early adolescents: Associations with depressive symptoms and health-related behaviors. Journal of Psychopathology and Behavioral Assessment, 48(3), Article 44. https://doi.org/10.1007/s10862-026-10306-3
- van den Eijnden, R. J. J. M., Lemmens, J. S., & Valkenburg, P. M. (2016). The Social Media Disorder Scale. Computers in Human Behavior, 61, 478–487. https://doi.org/10.1016/j.chb.2016.03.038
- World Health Organization. (2020, October 22). Addictive behaviours: Gaming disorder. https://www.who.int/news-room/questions-and-answers/item/addictive-behaviours-gaming-disorder
- World Health Organization. (2024). Clinical descriptions and diagnostic requirements for ICD-11 mental, behavioural and neurodevelopmental disorders. https://www.who.int/publications/i/item/9789240077263
Social Media Use: Construct Definition and Platform Heterogeneity
Social media describes a set of services rather than a single behavioral exposure. Messaging friends, scrolling feeds, posting creative work, following political news, participating in support groups, and engaging in monetized content all involve different motivational and social mechanisms. A general label of 'social-media addiction' risks combining heterogeneous activities and platform architectures. Investigators must define whether the target is an activity, a particular application, engagement with recommendation systems, or a cross-platform pattern of impaired control.
The 2025 proposal by Moretta and Wegmann Moretta, T., & Wegmann, E., 2025 organizes possible social-media-use disorder criteria around impairment, control, priorities, craving, and context-specific digital experiences. It is an explicit proposal, not an official addition to DSM-5-TR or ICD-11. The authors acknowledge the limited clinical-population validation available for the suggested criteria. This difference between proposing a clinically coherent construct and validating an independent disorder is central to interpreting the paper.
The scientifically productive question asks which forms of use, in which circumstances, generate durable impairment. Design features such as endless scrolling, notifications, public feedback metrics, algorithmically curated feeds, and monetization can reinforce interaction, but exposure to a feature is not proof of disorder. Researchers should examine the interaction of individual vulnerabilities and choice architecture, including whether a platform design change reduces harm independently of individual symptom scores.