The research was anchored in the Interaction of Person-Affect-Cognition-Execution (I-PACE) model, a theoretical framework for addictive behaviors that distinguishes between two kinds of engagement with a rewarding activity. Gratification-driven engagement reflects positive reinforcement: people play because the activity itself is immersive, meaningful, or tied to how they see themselves. Compensation-driven engagement, by contrast, reflects negative reinforcement: people play to escape distress, relieve stress, or satisfy internalized pressures. The team drew on the Comprehensive Gaming Motivation Inventory, which measures five motives within its immersion and escapism dimension—Coping, Escape, Fantasy, Identity, and Introjected Regulation. Based on the content of the questionnaire items, the researchers grouped Fantasy and Identity as gratification-oriented motives, and Coping, Escape, and Introjected Regulation as compensation-oriented ones.

Most previous studies linking gaming motivations to gaming disorder have been cross-sectional, capturing a single moment in time and therefore unable to establish which comes first. Longitudinal investigations remain relatively rare, yet they are essential for determining the temporal ordering of associations—though, as the authors emphasize, temporal precedence alone cannot establish causation. The research team had previously conducted a three-wave longitudinal study of emerging adults, in which compensation-oriented motives, particularly coping and escape, emerged as the primary longitudinal predictors of gaming disorder risk, while gratification-oriented motives did not predict later symptoms. Adolescence, however, involves ongoing and heterogeneous changes in reward processing, cognitive control, and goal-directed learning, raising the question of whether adult-derived models of addiction apply straightforwardly to younger populations.

To find out, the researchers recruited students aged 14 to 18 from a single secondary school, with the average participant aged 15.45 years. The final longitudinal sample included 71 girls, 42 boys, one non-binary participant, and six students who preferred not to disclose their gender. Eligibility required playing video games for at least one hour per week. Participants completed online questionnaires during class under supervision, passing two attention checks before their data were retained. To enable anonymous matching across waves, participants generated identification codes from randomly selected personal information. The first wave began in December 2024, with the second following three months later.

The path from initial recruitment to the final sample was substantial. Of 320 students who completed the first survey, 260 provided valid data after exclusions for ineligibility, failed attention checks, or invalid codes. At the second wave, 248 students participated, of whom 215 provided valid responses. Only 120 participants could be reliably matched across both waves, because matching relied entirely on self-generated anonymous codes—the researchers could not distinguish true attrition from cases where a participant simply entered a different code the second time. Comparisons between retained and non-retained participants showed no significant differences in age, gender, baseline gaming disorder symptoms, or four of the five motives, although retained participants reported somewhat higher baseline coping motives, a small difference the authors flag as a potential selection consideration.

Gaming disorder risk was measured with the Internet Gaming Disorder Scale Short Form, a nine-item questionnaire reflecting the DSM-5 criteria for internet gaming disorder, rated on a five-point scale from never to very often. The researchers stress that this instrument served as a screening indicator rather than a clinical diagnosis, and notably, no participant in the sample reached the research-based cut-off score of 36 or higher for disordered gaming at either wave. Gaming motivation was assessed with the Gaming Motivation Inventory, rated on a seven-point scale. Internal consistency of the scales was generally satisfactory across both waves, with one exception: the introjected regulation subscale showed weaker reliability at the first measurement.

Statistically, the team employed cross-lagged panel analysis with item-parceling methodology, aggregating questionnaire items into parcels used as indicators of latent constructs. This approach was necessary because a traditional two-wave cross-lagged model using scale mean scores would be statistically saturated and produce no fit indices. Each motive was analyzed in a separate model paired with gaming disorder risk, and reported values represent averages across 100 model replications. At both time points, all five motives correlated positively and significantly with gaming disorder risk, with the strongest concurrent associations observed for introjected regulation and fantasy—confirming, at the correlational level, the robust links between immersion and escapism motives and problematic gaming reported in earlier cross-sectional research.

The cross-lagged results told a more differentiated story. Supporting the gratification hypothesis, higher fantasy and identity motives at the first wave each predicted higher gaming disorder risk at the second wave, after accounting for baseline symptoms. The reverse path—from gaming disorder risk to later fantasy—did not reach statistical significance. Intriguingly, gaming disorder risk at the first wave predicted a subsequent decline in identity motives, meaning that adolescents with more severe symptoms later endorsed identity-related reasons for gaming less strongly. The authors caution that this finding concerns the specific questionnaire subscale, which measures whether gaming feels personally meaningful or identity-related, not general adolescent identity development, and that the underlying psychological processes were not directly assessed. The compensation hypothesis received partial support: gaming disorder risk prospectively predicted increases in coping and introjected regulation, but no significant cross-lagged association emerged for escape in either direction.

The null finding for escape is particularly noteworthy. The researchers suggest that escape may operate differently from other compensation-oriented motives—possibly being more sensitive to short-term fluctuations in distress, which could make its association with gaming disorder symptoms harder to detect over a three-month interval. They stress that this interpretation remains tentative and requires testing with more frequent measurement occasions. More broadly, the pattern is consistent with the possibility that among adolescents at higher risk, gaming becomes more strongly linked to coping with distress or responding to internalized pressures, echoing models in which compensatory motives grow more salient alongside increasing disorder risk. However, the authors are careful to note that the study does not demonstrate changes in goal-means relationships, nor that these motives emerge as consequences of gaming disorder itself.

The limitations of the design temper the conclusions considerably. With only two measurement occasions, the cross-lagged models cannot separate stable between-person differences from within-person processes, meaning the observed paths should not be read as evidence of individual-level change. Temporal precedence within the three-month window does not establish causality, since the associations may reflect stable individual differences, unmeasured confounding, or processes not represented in the models. All data were self-reported, the sample came from a single school in one national context, classroom-level clustering could not be assessed, and no covariate-adjusted analyses were feasible given the sample size. The authors call for future research with additional waves, larger and better-tracked samples, behavioral indicators such as playtime logs, and cross-cultural comparisons. If the motive-specific patterns are replicated, they could inform school-based prevention: fantasy and identity motives might serve as early risk indicators, while coping and introjected regulation could become targets for interventions aimed at strengthening emotion-regulation skills and alternative coping strategies.

Subject of Research: Longitudinal associations between gaming motivations and gaming disorder symptoms in adolescents

Article Title: Gratification vs. compensation: longitudinal evidence for the shifting motivations behind adolescent gaming disorder

Article References: Starzec, S., Zajas, A., Michałkiewicz, N., Kiszka, P., Kowalska, L., & Strojny, P. (2026). Gratification vs. compensation: longitudinal evidence for the shifting motivations behind adolescent gaming disorder. Frontiers in Psychiatry, 17, Article 1938938. https://doi.org/10.3389/fpsyt.2026.1938938

Image Credits: AI Generated

DOI: 10.3389/fpsyt.2026.1938938

Keywords: gaming disorder, adolescents, gaming motivation, I-PACE model, longitudinal study, cross-lagged panel analysis, gratification, compensation, fantasy motives, identity motives, coping, Frontiers in Psychiatry

The research, led by Yuanyuan Chen of Shaanxi University of Technology and Yongquan Huo of Shaanxi Normal University, together with colleagues at Anhui University of Chinese Medicine, Xi’an University, Tianshui Normal University, and Shaanxi University of Technology, was published in the International Journal of Mental Health and Addiction. The team recruited 1,034 participants aged 18 to 26 years from two universities in Xi’an and Haikou, China, and asked them to complete a detailed questionnaire measuring social anxiety, perceptions of their AI chatbots, emotional bonds with the systems, executive-function difficulties, and patterns of problematic use of conversational AI, abbreviated PUCAI in the study.

The headline finding is stark: social anxiety positively predicted problematic use of conversational AI. In other words, the more socially anxious a young adult reported being, the more likely they were to show signs of problematic engagement with chatbots — patterns resembling the loss of control, preoccupation, and continued use despite negative consequences that researchers associate with behavioral addictions. The result echoes earlier work, including a 2023 study in Computers in Human Behavior by Hu, Mao, and Kim, which found that loneliness, rumination, and mind perception linked social anxiety to problematic chatbot use, and a 2023 study by Wei and colleagues showing that adolescents’ social anxiety can trigger problematic smartphone use.

But the new study goes further by mapping the internal machinery of that relationship. Using statistical mediation analysis, the researchers found that three variables operated in parallel between social anxiety and PUCAI: perceived anthropomorphism, emotional attachment, and executive-function difficulties. Perceived anthropomorphism refers to the tendency to attribute human-like mental states, intentions, and personality to non-human agents — a phenomenon famously theorized by Epley, Waytz, and Cacioppo in their 2007 three-factor theory of anthropomorphism. When socially anxious users perceive a chatbot as more human-like, the machine becomes a safer social partner than any person, one that never judges, interrupts, or walks away.

Emotional attachment, the second mediator, describes the affective bond users develop with the system — a construct borrowed from consumer psychology, where Park and colleagues distinguished attachment from mere attitude toward a brand. In the context of conversational AI, attachment means the chatbot becomes a source of comfort and connection, something the user turns to in distress. Qualitative research on the social chatbot Replika, including a 2024 grounded theory analysis by Laestadius and colleagues published in New Media & Society, has documented how emotional dependence on such systems can produce real psychological harms — users described the bots as simultaneously too human and not human enough, a paradox that left them feeling both connected and bereft.

The third mediator, executive-function difficulties, brings the brain’s cognitive control system into the picture. Executive functions — working memory, inhibition, and cognitive flexibility — are orchestrated largely by prefrontal cortical networks, as Menon and D’Esposito outlined in Neuropsychopharmacology in 2022. These capacities mature gradually from adolescence into adulthood, following what Tervo-Clemmens and colleagues described in Nature Communications in 2023 as a canonical trajectory of executive function maturation. Young adults aged 18 to 26 sit squarely within this window of ongoing maturation, which may make them particularly vulnerable when the pull of an always-available, endlessly patient conversational partner collides with still-developing self-regulation. The study found that social anxiety predicted greater executive-function difficulties, which in turn predicted more problematic AI use — consistent with the I-PACE model’s insistence that addiction-like behaviors emerge from the interaction of personal predispositions, affective states, cognitive processes, and execution deficits.

That theoretical framing is central to the paper. The Interaction of Person-Affect-Cognition-Execution model, developed by Matthias Brand and colleagues and updated in Neuroscience and Biobehavioral Reviews in 2019, holds that addictive behaviors are not simply responses to a substance or platform but the product of a person’s vulnerabilities meeting affective and cognitive processes that gradually narrow and intensify until they dominate behavior. The new findings fit this architecture precisely: social anxiety supplies the person-level vulnerability, anthropomorphism and attachment supply the affective-cognitive engine, and executive-function difficulties supply the execution failure that turns heavy use into uncontrolled use.

The study also drew on Conservation of Resources theory, proposed by Stevan Hobfoll and elaborated by Shelley Taylor and colleagues, which holds that psychological well-being depends on maintaining valued internal and external resources — self-esteem, optimism, social support, a sense of mastery. Stress arises when those resources are threatened or lost. From this perspective, socially anxious individuals enter interactions with a depleted resource pool, and a chatbot that demands no social resources at all becomes an attractive refuge. Crucially, the researchers found that psychosocial resources moderated two of the key pathways: the associations of social anxiety with perceived anthropomorphism and with emotional attachment. In participants with richer psychosocial resources — greater perceived social support, psychological capital, and resilience — the link between social anxiety and seeing the chatbot as human, and the link between social anxiety and becoming attached to it, were weaker. Resources, in other words, act as a psychological buffer, dampening the tendency of social anxiety to convert into machine-directed intimacy.

This moderating effect carries real weight beyond theory. Taylor and colleagues have shown, in neuroimaging work published in the Journal of Personality and Social Psychology in 2008, that psychosocial resources literally moderate the brain’s cortisol stress responses, engaging prefrontal circuitry that down-regulates threat systems. Applied to the chatbot context, the message is that the same social anxiety can lead down very different paths depending on the resources a person carries: those with strong support networks and psychological reserves may use AI lightly and instrumentally, while those without may slide toward anthropomorphizing, attaching, and losing control. The findings also align with Chinese research on left-behind children by Fan and colleagues, which showed that psychosocial resources buffer the effects of adversity on social adjustment through similar moderated-mediation logic.

The practical implications are immediate. The authors suggest that interventions aimed at reducing problematic conversational AI use among socially anxious young adults should not simply target screen time. Instead, they should build psychosocial resources — social support, coping skills, psychological capital — and strengthen executive functions, while also addressing the anthropomorphism and attachment processes that make chatbots feel like companions. The timing could hardly be more urgent: China’s Internet Network Information Center reported in its 57th statistical survey that hundreds of millions of Chinese users now navigate an internet saturated with generative AI, and chatbots such as ChatGPT, Replika, and their Chinese counterparts are reaching ever younger audiences. At the same time, the field remains contested — researchers such as Ciudad-Fernández, von Hammerstein, and Billieux have argued in Addictive Behaviors that people are not becoming AIholic and that the ChatGPT addiction construct needs scrutiny. This study does not claim that chatbots are inherently addictive; it claims something more nuanced — that for a specific, identifiable group of vulnerable users, particular psychological pathways make problematic use likely, and that those pathways can be measured, understood, and interrupted. As conversational AI becomes the default interface between young people and the digital world, understanding who is at risk, and why, may prove one of the most consequential questions in mental health research.

Subject of Research: Psychology

Article Title: Social Anxiety and Problematic Use of Conversational AI: I-PACE Model and Conservation of Resources Perspectives

Article References: Chen, Y., Wang, M., Bai, Y., Yuan, S., Tian, W., & Huo, Y. (2026). Social Anxiety and Problematic Use of Conversational AI: I-PACE Model and Conservation of Resources Perspectives. International Journal of Mental Health and Addiction. https://doi.org/10.1007/s11469-026-01750-z

Image Credits: AI Generated

DOI: 10.1007/s11469-026-01750-z

Keywords: social anxiety, conversational AI, chatbots, problematic AI use, I-PACE model, conservation of resources, executive function, emotional attachment, anthropomorphism, mental health, Social, Anxiety

The study, led by Lei Yu of the Third People’s Hospital of Qujing with colleagues including Wenzhi Wu of Beijing Normal University, followed 1,474 adolescents across three annual assessments beginning in October 2021 and ending in October 2023. Participants, whose average age at baseline was about 15.5 years and nearly 60 percent of whom were female, were drawn from 27 classrooms in two public secondary schools in Southwest China. At each wave, the researchers measured three constructs: perceived stress, core self-evaluations, and problematic smartphone use. Perceived stress was assessed with the 10-item Perceived Stress Scale, which captures the appraisal that environmental demands exceed one’s coping resources rather than the mere presence of demands. Core self-evaluations, measured with the 10-item Core Self-Evaluation Scale, represent a higher-order judgment of one’s own worth, competence, sense of control, and emotional stability, integrating self-esteem, generalized self-efficacy, locus of control, and emotional stability into a single psychological resource.

Problematic smartphone use, the third pillar of the study, was measured with a 22-item scale covering withdrawal-like behavior, salience, social comfort, negative effects, and patterns of application use. The construct matters because it is distinct from heavy use. Frequent smartphone use is not inherently maladaptive; the developmental concern arises when use becomes compulsive or poorly controlled and begins to interfere with sleep, learning, emotional functioning, or offline relationships. Reviews have linked problematic smartphone use to poorer mental health and sleep outcomes in young people, but the direction of these associations and the mechanisms behind them have remained unsettled, largely because most studies are cross-sectional snapshots that cannot establish temporal ordering.

The methodological innovation at the heart of the new study is the random-intercept cross-lagged panel model, or RI-CLPM. Conventional cross-lagged panel models, which compare variables across repeated waves, have a well-documented weakness: they can blend stable between-person differences with genuine within-person change. In adolescence this distinction is critical. Some teenagers may be chronically more stressed, lower in self-evaluation, and more prone to problematic phone use because of relatively stable factors such as temperament, family context, academic environment, or peer experiences. Those stable differences are meaningful for identifying risk groups, but they are not the same as dynamic processes unfolding inside an individual. The RI-CLPM solves this by estimating a random intercept for each person on each variable, capturing that adolescent’s typical level, and then asking whether deviations from that own typical level predict later deviations. In other words, it asks whether a teenager who is more stressed than usual this year reports lower self-evaluation next year, and whether that dip forecasts more problematic phone use the year after.

The data showed that the separation was necessary. Intraclass correlations indicated that stable between-person differences accounted for roughly half or more of the total variance in all three constructs, between 52 and 59 percent, with the remainder reflecting within-person fluctuation. At the between-person level, the correlations were strong and unsurprising: adolescents who were generally lower in core self-evaluations tended to report more perceived stress and more problematic smartphone use, and generally more stressed adolescents reported more problematic use. The random intercepts for self-evaluations and stress were correlated at about negative 0.84, an unusually strong trait-level association. But these chronic risk profiles, the researchers emphasize, cannot be treated as evidence of within-person change.

The within-person findings were the study’s core contribution. Core self-evaluations and perceived stress were reciprocally and negatively linked over time. When adolescents reported self-evaluations higher than their own usual level, they subsequently reported lower perceived stress, with a standardized cross-lagged coefficient around negative 0.15. When they reported stress higher than their own usual level, they subsequently reported lower core self-evaluations, with a coefficient around negative 0.12. By the effect-size benchmarks proposed for cross-lagged designs, these associations fall in the large range. The pattern fits appraisal theories of stress, which hold that demands become stressful when coping resources are judged insufficient, and resource-based accounts in which personal resources both protect against stress and are depleted by sustained demands. Higher-than-usual self-belief may buffer the impact of academic competition, peer conflict, and family expectations; repeated experiences of overload, failure, or uncontrollability may in turn erode a teenager’s sense of competence and control.

The pathway to the phone ran through this resource, not around it. Higher-than-usual core self-evaluations predicted lower subsequent problematic smartphone use, a significant prospective association with a coefficient around negative 0.13. Perceived stress, by contrast, showed no direct predictive path to later problematic use once stable differences and self-evaluations were modeled. Yet a longitudinal indirect association from stress to problematic smartphone use through core self-evaluations was statistically different from zero, estimated with 5,000 bootstrap samples and bias-corrected confidence intervals. The reverse sequence, in which problematic use would erode self-evaluations and later raise stress, was not supported. This asymmetry gives the model a sharper developmental interpretation: at the annual time scale examined here, the road from stress to the smartphone passes through the self. The finding is consistent with compensatory internet-use theory and the I-PACE model, both of which frame problematic digital behavior as emerging from interactions among affective states, person-level vulnerabilities, cognitive appraisals, and executive control, rather than from stress alone.

The authors are careful about what the design can and cannot claim. Because the study is observational, the temporally ordered indirect associations cannot be read as causal mediation, and the hypotheses and analytic plan were not preregistered. The researchers also stress-tested their conclusions. The results held under 200 multiple imputations of missing data, under classroom-cluster-robust standard errors accounting for nesting within 27 classrooms, and under partial scalar measurement-invariance specifications. The within-person pathways did not differ significantly between boys and girls, though the authors caution that this invariance concerns the path coefficients only, not mean levels, stress exposure, or patterns of use. One measurement caveat deserves attention: the Perceived Stress Scale showed only modest internal consistency in this sample, with reliability coefficients between roughly 0.66 and 0.72, which may have attenuated associations involving stress and warrants replication with alternative measures.

The practical implications are framed as hypotheses for future work rather than prescriptions. The prospective association between core self-evaluations and later problematic smartphone use suggests that self-evaluative resources may be worth examining as a candidate mechanism in experimental and intervention research, for example by testing whether programs that strengthen perceived competence, control, coping confidence, or emotion regulation produce subsequent changes in smartphone behavior. But until such evidence exists, core self-evaluations should be viewed as a potentially relevant psychological process, not an established intervention target. The authors also note limitations of scale and scope: all constructs were self-reported, the one-year lag may not match the faster rhythms at which stress and phone use interact day to day, and the sample came from only two public secondary schools in Southwest China, limiting generalizability across regions and school types. Future studies combining self-reports with device-recorded behavior, sleep indicators, and daily diary or ecological momentary assessment could capture shorter-term dynamics that annual waves cannot.

What the study ultimately offers is a more conditional account of why some adolescents lose control of their smartphone habits. Stressful periods appear most relevant to problematic use when they are accompanied by declines in self-evaluative resources; a generalized stress-to-phone link was not supported. Core self-evaluations, often treated as a fixed background trait, emerge instead as a dynamic resource that fluctuates meaningfully around each adolescent’s typical level and carries developmental information across years. For a generation growing up with smartphones woven into school coordination, peer contact, and family communication, the message from this three-year dataset is subtle but consequential: the psychological resource that matters may not be how much stress a teenager faces, but how capable, in control, and emotionally stable that teenager feels while facing it.

Subject of Research: Within-person longitudinal relations among perceived stress, core self-evaluations, and problematic smartphone use in Chinese adolescents

Article Title: Within-person reciprocal relations among perceived stress, core self-evaluations, and problematic smartphone use in Chinese adolescents

Article References: Yu, L., Wu, W., Jia, Y., Chang, L., & Ji, J. (2026). Within-person reciprocal relations among perceived stress, core self-evaluations, and problematic smartphone use in Chinese adolescents. Current Psychology, 45(20), Article 1587. https://doi.org/10.1007/s12144-026-10123-7

Image Credits: AI Generated

DOI: 10.1007/s12144-026-10123-7

Keywords: adolescents, perceived stress, core self-evaluations, problematic smartphone use, random-intercept cross-lagged panel model, longitudinal study, self-esteem, compensatory internet-use theory, I-PACE model, China, mental health, self-regulation

The I-PACE framework, short for Interaction of Person-Affect-Cognition-Execution, was originally developed by Matthias Brand and colleagues to explain addictive behaviors ranging from internet-use disorders to gambling. The model proposes that a person’s characteristics, emotional states, and cognitive appraisals interact with executive functions to shape whether a behavior remains controlled or spirals into compulsion. In the context of generative AI, the researchers asked how seven psychological variables—self-efficacy, need for cognition, avoidance learning motivation, positive affect, emotion regulation, perceived usefulness, and cognitive absorption—related to dependence. The choice of variables reflects the model’s core logic: distal factors such as personality-like dispositions do not act directly on behavior, but instead work through more proximal mechanisms such as how useful a technology feels and how deeply it captures attention.

The study’s most striking result concerns cognitive absorption, a state of deep, total immersion in an activity in which time seems to dissolve and attention narrows to the task at hand. Among all the variables examined, cognitive absorption emerged as the strongest direct correlate of generative AI dependence. This finding matters because it reframes what absorption actually is in the AI context. Previous research on problematic internet use and short-form video consumption has often treated absorption as a symptom of addiction, a byproduct of the compulsive behavior itself. Sun and Ma’s analysis instead positions cognitive absorption as an antecedent—a precursor that comes before dependence and helps create it. In practical terms, the students most at risk may be those who find the conversational flow of AI tools mesmerizing, not merely those who use them most often.

Beyond the absorption-centered route, the study identified a second, distinctly different pathway to dependence: an effort-minimizing route that runs through perceived usefulness. Students who came to see generative AI as an efficient shortcut for academic work were more likely to develop dependent patterns of use. This pathway is subtler and arguably more insidious than the absorption route, because it does not involve losing oneself in the technology at all. Instead, it involves a cold calculation: the tool saves effort, the effort saved feels valuable, and gradually the student stops attempting tasks without it. The researchers describe these as two distinct routes to the same destination, and their coexistence helps explain why AI dependence looks so different from one student to another—some are drawn in by fascination, others by convenience.

The indirect pathways in the model reveal equally important nuances. Positive affect and emotion regulation were found to influence dependence only indirectly, operating through perceived usefulness. In other words, students who experience more positive emotions and who regulate their emotions well come to view AI as more useful, and it is that perception of usefulness—not the positive mood itself—that feeds dependence. This finding connects with the broaden-and-build theory of positive emotions, which holds that pleasant emotional states expand a person’s thought-action repertoire and build lasting resources. Here, however, those resources appear to be channeled into an increasingly favorable appraisal of AI tools, which can then tip into overreliance.

Perhaps the most counterintuitive results involve need for cognition, self-efficacy, and avoidance learning motivation. All three were related to dependence only indirectly, rather than exerting direct effects. Need for cognition—the tendency to enjoy and seek out effortful thinking—has long been considered protective against problematic technology use, since students who love thinking hard might be expected to resist shortcuts. The study’s structural model suggests that these dispositions still matter, but that their influence is transmitted through the proximal mechanisms of usefulness and absorption rather than acting on dependence directly. Similarly, avoidance learning motivation—studying not to master material but to escape negative outcomes such as failure or criticism—was tied to dependence only through indirect routes, consistent with avoidance motivation theory’s account of how escape-oriented goals shape technology choices.

Methodologically, the researchers used structural equation modeling, the standard technique for testing networks of direct and indirect relationships among latent psychological constructs. The survey sample of 1,029 students spanned both high school and university populations in China, a country where internet penetration and AI adoption have grown at extraordinary pace according to recent national statistics. The study received ethics approval from the Ethics Committee of Yangzhou University Medical College, and for participants under eighteen, the survey was administered with school approval alongside informed notification to students and their parents or guardians. The authors report no competing financial interests, and the work was supported by the National Natural Science Foundation of China and several provincial and municipal research funds.

The theoretical contribution of the study lies in how it refines the I-PACE model itself. By revealing utility-based indirect pathways, the research extends a framework built primarily around gaming, pornography, and social media into the domain of productivity-oriented AI tools, where the technology is valued not for pleasure but for performance. This distinction is crucial: classic behavioral addiction models emphasize hedonic gratification, yet generative AI dependence may often begin with a rational, even admirable desire to work more efficiently. The study also elevates cognitive absorption from symptom to antecedent, implying that interventions targeting absorption—such as encouraging deliberate pauses, reflective self-monitoring, and mindful engagement with AI outputs—might prevent dependence before it forms, rather than treating it after the fact.

The practical implications reach students, educators, and policymakers alike. For teachers, the two-pathway finding suggests that blanket restrictions on AI use may miss the mark: the student who is mesmerized by chatbot conversations and the student who quietly outsources every assignment need different kinds of guidance. For the first, interventions might focus on attention management and awareness of immersive states; for the second, on rebuilding the perceived value of independent effort and addressing the avoidance motivations that make shortcuts attractive. For policymakers drafting national AI-in-education strategies, the results argue for psychological literacy alongside technical literacy, so that students learn not only how to use AI but how their own minds respond to it.

The study also arrives amid a broader wave of research on AI and mental health. Related work has examined AI dependence among teachers, the role of academic stress and self-efficacy in problematic AI usage, links between AI reliance and critical thinking, and the emergence of deskilling in the generative AI era. International bodies including the OECD and UNDP have published reports on effective and humane uses of AI in education, while Stanford’s annual AI Index documents the technology’s accelerating diffusion. Against this backdrop, Sun and Ma’s contribution is a reminder that the question is no longer whether students will use generative AI, but which psychological currents will carry some of them from casual use into dependence—and how education systems can read those currents early enough to respond.

Subject of Research: Psychological mechanisms of student dependence on generative AI analyzed through the I-PACE behavioral addiction framework

Article Title: Are Students Dependent on Generative AI? Analyzing Psychological Mechanisms Through the I-PACE Framework

Article References: Sun, F., & Ma, R. (2026). Are Students Dependent on Generative AI? Analyzing Psychological Mechanisms Through the I-PACE Framework. International Journal of Mental Health and Addiction. https://doi.org/10.1007/s11469-026-01740-1

Image Credits: AI Generated

DOI: 10.1007/s11469-026-01740-1

Keywords: generative AI dependence, I-PACE model, cognitive absorption, perceived usefulness, emotion regulation, need for cognition, self-efficacy, avoidance motivation, students, behavioral addiction, educational technology, China

The study, led by Sergio Hidalgo-Fuentes of the Universitat de València together with Isabel Martínez-Álvarez and Elisa González-Pérez of the Universidad a Distancia de Madrid, pooled data from 14 studies encompassing 15 independent samples and a remarkable 79,760 participants. Following PRISMA guidelines and registered in PROSPERO, the researchers systematically searched PsycINFO, Scopus, PubMed, and Web of Science in February 2026. Their central finding: resilience and IGD symptomatology show a statistically significant, moderately negative correlation of r = −.24 (95% CI [−.29, −.20]). In plain terms, individuals who report higher resilience consistently report fewer symptoms of problematic gaming—and every single study included in the analysis reported a negative correlation, with effect sizes ranging from −.11 to −.46.

Resilience, broadly defined as the capacity to maintain or regain psychological equilibrium in the face of adversity, has long been recognized as a protective factor against addictive behaviours ranging from substance abuse to problematic internet use. But the construct itself is anything but simple. The literature distinguishes three complementary perspectives: resilience as a relatively stable personality trait marked by perseverance, emotional stability, and optimism; resilience as a dynamic, context-dependent process emerging from ongoing person–environment interactions; and resilience as a successful outcome despite significant adversity. These conceptual differences matter, because they shape how resilience is measured—and, as this meta-analysis reveals, how it is measured turns out to influence the size of its apparent relationship with IGD.

The measurement question proved to be the study’s most striking methodological insight. Among all the moderating variables examined, only the resilience assessment instrument significantly influenced the strength of the association, explaining 56.23% of the between-study variance. Instruments such as the Connor–Davidson Resilience Scale conceptualize resilience as a multidimensional trait reflecting personal competence and stress tolerance, while the Brief Resilience Scale focuses narrowly on the ability to “bounce back” from stressful situations. Because these tools capture partially distinct facets of the construct, the authors argue, they should not be treated as interchangeable—a caution with real consequences for how future studies of gaming disorder are designed and compared.

The theoretical framework underlying the findings is the I-PACE model—Interaction of Person–Affect–Cognition–Execution—one of the most influential accounts of behavioural addictions. According to this model, technology-related disorders arise from the interplay of personal vulnerabilities, dysfunctional affective states, biased cognitive processes, and deficits in executive functioning. Resilience, on this view, operates as a transversal personal resource: it buffers the emotional impact of stress, moderates gaming-related expectations of escape and reward, and supports inhibitory control, thereby reducing the likelihood that gaming becomes a vehicle for emotional regulation. This is especially pertinent for adolescents, whose heightened neuropsychological sensitivity to immediate rewards and immature inhibitory control help explain their overrepresentation in IGD epidemiology.

Complementary frameworks reinforce the same picture. The Self-Medication Hypothesis proposes that addictive behaviours emerge as maladaptive strategies for regulating psychological distress; from this angle, individuals with lower resilience may turn to gaming precisely because they lack more adaptive coping resources. Neuroscientific work by Ersche and colleagues suggests that dysfunction in self-regulation circuitry may underlie vulnerability to addiction, with resilient individuals partially compensating for reduced inhibitory control. Consistent with these accounts, prior studies have linked low resilience not only to greater IGD risk but also to more severe depression, stress, and difficulties with emotional regulation among young people.

The meta-analysis was not without statistical caveats. Heterogeneity was substantial—Cochran’s Q reached 144.09 with an I² of 95.20%—indicating that while the direction of the association was universal, its magnitude varied widely across studies. The prediction interval of −.40 to −.07 suggests future studies could observe effects anywhere in that range. Yet sensitivity analyses confirmed robustness: a leave-one-out procedure showed pooled estimates fluctuating only between r = −.24 and r = −.25 regardless of which study was excluded. Publication bias was also assessed through funnel plots, Egger’s regression test, and the Begg and Mazumdar rank correlation test; despite slight visual asymmetry, both statistical tests were non-significant, allowing the authors to reasonably dismiss publication bias.

The evidence base itself has clear demographic contours. Most included studies were conducted in Asia (57.14%), followed by Middle Eastern countries (28.57%) and Europe (14.29%), with all employing cross-sectional designs published between 2019 and 2025. Participants were predominantly adolescents and young adults, with mean ages ranging from 12.84 to 25.57 years, and sample sizes spanning from 71 to an extraordinary 50,855 individuals. The Internet Gaming Disorder Scale–Short Form dominated IGD assessment. Older adults, clinical populations, and non-student gamers remain markedly underrepresented, and the authors stress that the pooled estimate should be read as an average association rather than a uniform effect across all populations. Language restrictions to English and Spanish and exclusive reliance on self-report measures—which may inflate associations through shared method variance and social desirability—add further interpretive caution.

Perhaps the most important limitation is the field’s oldest problem: correlation is not causation. Because every included study was cross-sectional, the meta-analysis cannot determine whether low resilience drives gaming disorder, whether persistent gaming erodes resilience through sleep disruption, social withdrawal, and disengagement from adaptive activities, or whether both dynamics operate simultaneously. The authors call explicitly for longitudinal and experimental research to clarify temporal ordering and to test whether resilience-enhancing interventions actually reduce IGD symptoms. Prior meta-analytic evidence shows resilience interventions can improve mental health and adaptive coping among students, but whether those benefits translate into reduced problematic gaming remains an open question.

Even with those constraints, the practical implications are significant. Effects of this magnitude are typical for psychological risk and protective factors and can be meaningfully consequential at the population level. The findings argue for embedding resilience-building strategies—emotion regulation training, adaptive coping skills, stress management—into school-based and community prevention programmes targeting adolescents and young adults, the demographic where IGD prevalence climbs to 4–10% or higher and where East Asian estimates reach roughly 6%. For a disorder formally recognized by both major psychiatric classification systems, this synthesis provides researchers and clinicians with something they previously lacked: a rigorously quantified, robust, and bias-checked estimate of how psychological resilience relates to problematic gaming, along with a clear roadmap for the longitudinal, cross-cultural, and clinical work that must follow before causality can be claimed.

Subject of Research: The relationship between psychological resilience and internet gaming disorder, quantified through meta-analysis.

Article Title: The Association Between Internet Gaming Disorder and Resilience: A Meta-Analysis

Article References: Hidalgo-Fuentes, S., Martínez-Álvarez, I., & González-Pérez, E. (2026). The Association Between Internet Gaming Disorder and Resilience: A Meta-Analysis. International Journal of Mental Health and Addiction. https://doi.org/10.1007/s11469-026-01715-2

Image Credits: AI Generated

DOI: 10.1007/s11469-026-01715-2

Keywords: internet gaming disorder, resilience, meta-analysis, behavioural addiction, adolescents, mental health, I-PACE model, gaming disorder, coping strategies, prevention, psychology, self-regulation