New Framework Models Problematic Short-Form Video Use

A structured integrative conceptual review published in Frontiers in Psychology proposes a new theoretical framework to explain how problematic short-form video use may emerge and become self-maintaining. The study, led by researchers from Hebei GEO University and Hebei Normal University, introduces the Reinforcement–Narrowing–Shift (RNS) model, which focuses on the dynamic interaction between platform design and user behavior. Unlike previous accounts that often emphasize addictive-like symptoms or individual self-control, this framework highlights the exposure environment that recommendation systems continually update. The authors argue that understanding this recursive process is essential for developing effective interventions and research strategies.

The RNS framework posits that interactions between specific platform affordances and user vulnerabilities can reinforce repeated engagement. These affordances include auto play, infinite scroll, full-screen presentation, rapid switching, and fine-grained behavioral feedback. The review suggests that these design features do not act in isolation but create a media environment where content is short, emotionally salient, personalized, and continuously available. This environment couples interface design, behavioral reinforcement, and algorithmic selection, making continuation easier than deliberate stopping for some users.

Central to the model is the concept of informational narrowing, which the authors distinguish from related terms like filter bubbles and echo chambers. While filter bubbles refer to algorithmic personalization at the platform level, and echo chambers describe social network structures, informational narrowing refers to a dynamic process where informational diversity progressively contracts across multiple dimensions. The framework identifies four distinct dimensions of this narrowing: content-level, affective, cognitive-demand, and temporal-horizon. These dimensions are not necessarily linked to general cognitive decline but rather to the specific structure of the content and use episodes.

The review explains that under specific boundary conditions, repeated engagement may contribute to this multidimensional narrowing. These conditions include engagement-dominant optimization, stable behavioral signals, weak diversity constraints, and predominantly passive consumption. When these factors align, the feed may converge on highly stimulating or familiar content, delivered in rapid succession. This convergence can lead to a state where the user’s behavioral signals and the algorithmic output become strongly coupled, creating a self-reinforcing loop that is difficult to interrupt without external or deliberate internal intervention.

Three bridging mechanisms are proposed to explain how narrowed exposure translates into cognitive-affective shifts. The first is processing fluency and metacognitive erosion, where predictable feeds make continuation feel effortless, potentially reducing the user’s ability to monitor their own behavior. The second is affective habituation and escalation, where repeated exposure to similar emotional tones leads to a need for more intense content to achieve the same level of stimulation. The third is attentional capture and automaticity, where platform-related cues trigger automatic engagement that bypasses deliberate decision-making. These mechanisms may reinforce one another, making the problematic use pattern more persistent.

The framework is presented as a recursive process, meaning that later cognitive-affective shifts and renewed engagement can reactivate earlier reinforcement processes. The authors outline three analytical stages: initiation and reinforcement, consolidation through narrowing, and maintenance and escalation. In the initiation stage, engagement is supported by novelty and exploration. In the consolidation stage, the feed becomes more predictable and habit-driven. In the maintenance stage, the pattern becomes entrenched, characterized by persistence and reduced flexibility. These stages are not diagnostic categories but analytical distinctions that may overlap or recur differently across users.

To test the framework, the authors developed nine falsifiable propositions organized into three research modules. Module A involves a repeated-measures digital-trace panel, combining app-opening experience sampling, passive session logs, and feed sampling. Module B is a controlled feed experiment that manipulates diversity, predictability, or cue exposure. Module C is a comparative extension across different platforms or cultures. These modules are designed to address specific hypotheses, such as whether cue-triggered use predicts gaps between intended and actual disengagement, or whether diversity-aware recommendation weakens the association between repeated engagement and exposure narrowing.

The review also discusses the role of cultural and institutional moderators in the RNS cycle. Cultural contexts may shape why users enter the problematic use cycle and how they interpret narrowing. For instance, in contexts emphasizing social harmony, narrowing toward group-congruent content may be experienced as social participation rather than restriction. Institutional environments, including regulatory frameworks and educational systems, can also influence the intensity of the cycle and the availability of exit pathways. The authors note that direct evidence on these moderators is limited, but they provide a theoretically grounded agenda for future cross-cultural and cross-institutional research.

Intervention implications are derived from the mechanism-matched multilevel framework. Platform-level strategies aim to weaken reinforcement before narrowing stabilizes, using tools like autoplay limits and diversity-aware ranking. Educational strategies target the transition from narrowing to cognitive-affective shift by strengthening metacognitive monitoring and algorithmic literacy. User-level strategies focus on restoring reflective interruption at moments of strain, using techniques like implementation intentions and emotion-regulation training. The authors emphasize that single-level interventions are unlikely to fully disrupt the recursive cycle, and that coordinated multilevel approaches are necessary for effective prevention and reduction of problematic use.

The study acknowledges several limitations, including the fact that the causal ordering and full recursive structure of the RNS framework remain untested. Most available evidence is cross-sectional, self-reported, or limited to isolated components, and reverse causality is plausible. The authors caution that the framework should be treated as a falsifiable research agenda rather than a validated diagnostic or causal model. They recommend future studies use repeated-measures protocols and targeted feed experiments to test the specific linkages proposed in the model. This approach will help determine whether the RNS framework provides explanatory value beyond existing individual-level factors.

In refining the definition of problematic short-form video use, the review clarifies that the phenomenon represents dysregulated media engagement rather than simply high-frequency consumption or a formal psychiatric disorder. The authors distinguish their model from established paradigms such as the Interaction of Person-Affect-Cognition-Execution model. While existing frameworks emphasize predisposing individual characteristics, cue-reactivity, and general executive dysregulation, the recursive process model uniquely treats the informational environment as an endogenous and measurable variable. Under this analytical formulation, a user’s behavior at an initial time point actively transforms the recommendation feed encountered at the subsequent interval, which subsequently alters attention, emotional response, and later actions. Consequently, within-person feed convergence is expected to predict downstream monitoring errors and disengagement difficulties even after accounting for baseline vulnerability and total screen time.

The conceptual synthesis was developed using a structured integrative methodology supported by a literature search completed on October 1, 2026, across PubMed, Crossref, and Google Scholar, alongside forward and backward citation chaining. Due to methodological variations across platforms, the authors noted the literature precludes pooled effect estimates or claims of exhaustive systematic review. Addressing empirical platform dynamics, the review notes findings from a one-year observational investigation of new short-video platform users. That research detected growing content homogeneity over time, but revealed that accurate personalization did not automatically trigger severe information cocoons, highlighting that narrow individual preferences substantially shape exposure trajectories. The authors stress that personalization is not synonymous with contraction, as exploration algorithms, active user searching, shifting interests, and external disruptions can maintain or restore exposure diversity.

Individual vulnerabilities also interact with platform mechanics to shape habitual consumption. While short-form video engagement often begins with intentional goals like relaxation or humor, recurrent use can turn fatigue, stress, or app icons into automatic triggers. Maladaptive patterns arise when brief audiovisual stimulation becomes the primary mechanism for affect regulation, reducing opportunities to practice slower coping responses. Specific personality profiles heighten this risk: impulsivity elevates the appeal of immediate gratification, boredom proneness intensifies attraction to fast transitions, neuroticism promotes platform use for mood repair, and lower conscientiousness increases routine interference with sleep, study, or work. Although causal directions remain unconfirmed, existing cross-sectional evidence associates intensive use with reduced attentional control, weaker self-regulated learning, lower learning motivation, and poorer perceived academic writing competence.

To operationalize the four dimensions of narrowing, the framework proposes concrete empirical metrics that distinguish feed characteristics from behavioral consequences. Content-level narrowing can be tracked through topic entropy, semantic distance, source diversity, and coverage indices. Affective narrowing can be quantified by examining valence and arousal distributions, emotion-category entropy, and within-session emotional transitions, separating emotional concentration from escalation. For experience-level features, cognitive-demand narrowing is indexed through linguistic complexity, argument structure, information density, and required viewing continuity, rather than inferred from generalized cognitive ability. Temporal-horizon narrowing is evaluated through transition density, the temporal distribution of successive videos, and the frequency of explicit stopping cues. Separating these structural properties from downstream consequences prevents researchers from conflating environmental feed architecture with user stopping failures or time-estimation distortions.

Beyond individual and platform mechanisms, institutional and public health contexts moderate how readily the cycle develops or resolves. Where public mental health support is accessible, affordable, and culturally accepted, individuals have greater opportunities to adopt alternative coping mechanisms before video viewing becomes an entrenched regulatory crutch. Conversely, limited mental health infrastructure can heighten dependence on immediate algorithmic distraction. When evaluating institutional interventions, the authors note that techniques adapted from gambling regulation, such as automated risk detection based on mismatches between realized behavior and reflective evaluation, could provide minimally invasive friction. However, they caution against pathologizing ordinary high-frequency engagement. Any automated risk detection deployed on short-form video platforms must incorporate informed consent, transparent operational thresholds, error monitoring, and independent oversight to safeguard privacy and avoid unwarranted clinical labeling.

Subject of Research: A structured integrative conceptual review proposing the Reinforcement–Narrowing–Shift framework for problematic short-form video use.

Article Title: Toward a recursive process model of problematic short-form video use: the Reinforcement–Narrowing–Shift framework

Article References: Ge, W., Hua, Y., Yuan, Y., Li, S., & Li, D. (2026). Toward a recursive process model of problematic short-form video use: the Reinforcement–Narrowing–Shift framework. Frontiers in Psychology, 17, Article 1871950. https://doi.org/10.3389/fpsyg.2026.1871950

Image Credits: AI Generated

DOI: 10.3389/fpsyg.2026.1871950

Keywords: short-form video, problematic use, algorithmic recommendation, informational narrowing, behavioral addiction, digital media, reinforcement learning, cognitive-affective shift, platform design, research framework, Toward, recursive