🧬 biology

Internet gaming disorder: dynamic functional connectivity and graph theory

This study reveals that Internet gaming disorder is characterized by a state-dependent redistribution of temporal brain network occupancy and specific topological disruptions in a transient connectivity state, with dynamic graph features in certain states successfully predicting symptom severity.

Original authors: Yichen Guo, Guohao Lu, Xinyu Wang, Wenjing Li, Longyao Ma, Yang Liu, Sha Zhou, Yong Zhang, Weijian Wang

Published 2026-08-20

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Yichen Guo, Guohao Lu, Xinyu Wang, Wenjing Li, Longyao Ma, Yang Liu, Sha Zhou, Yong Zhang, Weijian Wang

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The human brain is not a static machine that hums at a single, unchanging frequency. Instead, it is a dynamic landscape where different groups of neurons light up and fade away in shifting patterns, even when a person is simply resting with their eyes closed. Scientists have long used brain scans to map these connections, but for a long time, they treated the brain like a photograph, capturing a single, blurry average of activity over several minutes. This approach assumes that the brain's wiring stays the same throughout the scan, an idea that modern research increasingly doubts. We now know that the brain's connections fluctuate rapidly, reorganizing themselves into distinct, temporary states every few seconds. Understanding these fleeting moments is crucial for unraveling how the brain works, and more importantly, how it breaks down in conditions like addiction. When people become dependent on internet gaming, their brains undergo profound changes, but it has been unclear whether these changes are a constant, global disruption or something more specific that happens only during certain moments of mental activity.

A team of researchers at the First Affiliated Hospital of Zhengzhou University set out to look at this problem with a sharper lens. They studied 58 individuals diagnosed with internet gaming disorder and 56 healthy volunteers, all of whom underwent a resting-state functional magnetic resonance imaging scan. Instead of averaging the entire scan into one picture, the researchers broke the data down into tiny slices, looking at how brain regions connected to one another over short windows of time. They then used a computer method to group these fleeting patterns into four recurring "states" or configurations that the brain visited repeatedly. Think of these states as different moods or modes of operation the brain cycles through, each with its own unique map of connections. The team then asked a simple but profound question: do people with internet gaming disorder spend their time in these states differently than healthy people, and is the internal structure of these states different?

The results revealed a story that is far more nuanced than a simple "broken brain." The researchers found that the brains of people with internet gaming disorder did not look disorganized all the time. Instead, the disorder was characterized by a specific change in how the brain moved between these states and how it functioned when it was in one particular state. People with the disorder spent less time in two of the four states, specifically the ones where the brain's regions were most strongly connected and working together efficiently. They also left these highly connected states more quickly, staying there for shorter periods. Conversely, they spent more time in a state where the brain's connections were weaker and less integrated. This suggests that the addicted brain has a harder time settling into the efficient, coordinated modes of operation that are necessary for self-control and goal-directed thinking, preferring instead to linger in a more fragmented, less connected mode.

However, the most striking discovery was not just about how long the brain stayed in a state, but what happened inside that state when it did arrive. The researchers discovered that the disruption was confined to a single, specific state. In this one particular configuration, the brain networks of people with internet gaming disorder were significantly less efficient than those of healthy controls. The connections between different parts of the brain were weaker, and the pathways for information to travel across the network were longer and less direct. This meant that the brain was struggling to integrate information from different regions, a process that is essential for making decisions and inhibiting impulses. Crucially, this inefficiency was not present in the other three states the brain visited. The disorder did not cause a global collapse of brain organization; rather, it created a specific bottleneck that only appeared when the brain entered this particular mode of operation.

The study also found that this specific brain pattern was linked to the severity of the addiction. The researchers used the brain data to predict the scores of the participants on a standard test for internet addiction. They found that the more severe the addiction, the more pronounced the inefficiency was in that specific state. This connection held true even when the researchers tested their findings repeatedly to ensure they were not just a fluke. It suggests that the degree to which the brain fails to organize itself efficiently in this specific moment is a direct reflection of how deeply the person is affected by the disorder. The findings offer a new way to understand addiction, moving away from the idea of a permanently damaged brain toward a view of a brain that struggles to access or maintain its most efficient working modes.

While these findings provide a clearer picture of the neural mechanics of internet gaming disorder, the researchers are careful to note that this is a snapshot in time. The study was cross-sectional, meaning it compared two groups at a single point rather than following them over years. Therefore, it is not yet known whether these specific brain patterns are the cause of the addiction, a result of it, or a sign that the brain is struggling to cope with the symptoms. The participants were also predominantly male, and the study was conducted at a single location, which means the results may not apply to everyone. Nevertheless, the work demonstrates that the key to understanding complex behavioral disorders may lie in looking at the brain's moment-to-moment fluctuations rather than its average state. By identifying the precise moments when the brain's organization falters, scientists can begin to develop more targeted ways to help people regain control, potentially focusing on therapies that help the brain learn to enter and stay in those efficient, well-connected states.

Technical Summary: Internet Gaming Disorder: Dynamic Functional Connectivity and Graph Theory

Problem Statement
Internet Gaming Disorder (IGD) is recognized as a behavioral addiction characterized by weakened executive control, heightened reward sensitivity, and persistent craving. While resting-state functional connectivity (rsFC) studies have identified network-level disturbances in IGD, the prevailing approach relies on static, scan-wide averages. This methodology assumes connectivity is stationary, potentially obscuring the transient, structured fluctuations that occur over seconds to minutes. Furthermore, existing graph-theoretical analyses of IGD have yielded heterogeneous findings, likely because averaging connectivity over an entire scan dilutes state-specific topological disruptions. The central problem addressed is whether the neural signature of IGD is a pervasive, static deficit or a feature of specific, transient connectivity configurations, and whether state-resolved graph theory can better characterize network organization and predict symptom severity.

Methodology
The study analyzed resting-state fMRI data from 114 right-handed participants: 58 individuals diagnosed with IGD (meeting DSM-5 criteria and scoring ≥50 on the Internet Addiction Test, IAT) and 56 healthy controls (HC).

Data Preprocessing: Images were acquired on a 3T Siemens scanner (TR = 1000 ms, 400 volumes). Preprocessing included slice-timing correction, realignment, normalization, smoothing, and regression of motion parameters and nuisance signals.
Dynamic Functional Connectivity (dFC): Time-varying connectivity was estimated using a sliding-window approach (60-TR window, 2-TR step) on Dosenbach-160 regional time series, generating 168 windows per participant.
State Clustering: A k-means clustering algorithm (Lloyd's method) was applied to the concatenated window feature vectors. A four-state solution ( $K=4$ ) was selected based on a geometric elbow criterion, validated through repeated initializations, split-half resampling, and window-length stability analyses.
Graph-Theoretical Analysis: For each participant and state, mean connectivity matrices were computed. Graph metrics (clustering coefficient, characteristic path length, global/local efficiency, small-worldness) were calculated across a range of sparsity thresholds (0.03–0.40) and integrated as Area-Under-the-Curve (AUC) summaries.
Statistical Inference: Group differences in temporal metrics (fractional occupancy, mean dwell time, transitions) and graph metrics were assessed using General Linear Models (GLM) with covariates (age, sex, education, motion). Significance was determined using Benjamini-Hochberg False Discovery Rate (FDR) correction and Freedman-Lane residual permutation inference.
Prediction Modeling: Within the IGD group, state-specific graph features were used to predict IAT scores using Linear Support Vector Regression (SVR) with leave-one-out cross-validation (LOOCV) and robustness checks via repeated nested cross-validation.

Key Results

Temporal Dynamics: The IGD group exhibited a state-dependent redistribution of temporal occupancy. Specifically, IGD participants showed lower fractional occupancy and shorter mean dwell times in State 1 (strongly connected) and State 4 (moderately strongly connected), while showing higher fractional occupancy and longer dwell times in State 3 (weakly connected). No significant differences were found in State 2 or total transitions.
State-Specific Topological Disruption: Group differences in graph topology were confined exclusively to State 4. In this state, the IGD group demonstrated:
- Reduced clustering coefficient, local efficiency, normalized clustering, and small-worldness.
- Reduced global efficiency.
- Increased characteristic path length.
- These metrics indicated a simultaneous erosion of both functional segregation and integration, deviating from the efficient small-world architecture seen in controls. No significant topological differences were observed in States 1, 2, or 3.
Prediction of Symptom Severity: State-specific graph features successfully predicted IAT scores within the IGD group in State 2 ( $r=0.56$ ) and State 4 ( $r=0.57$ ) after FDR correction. Notably, State 4 was the only state showing both a group-level topological difference and a correlation with symptom severity. State 2 predicted severity despite showing no group-level topological difference. Robustness analyses (nested cross-validation) confirmed these findings with median $r$ values of 0.54 and 0.58, respectively.

Significance and Claims
The authors claim that these findings support a state-resolved, graph-based view of IGD, challenging the notion of a pervasive, scan-wide network deficit. The study posits that IGD is characterized by:

State-Specificity: The disorder's topological disruption is not global but confined to a single transient connectivity state (State 4), where the brain fails to maintain efficient small-world organization.
Temporal Redistribution: IGD involves a bias away from strongly integrated configurations (States 1 and 4) toward a weakly connected, default-like regime (State 3), potentially reflecting weakened executive control and reward-driven disengagement.
Clinical Relevance: The topology of State 4 scales with addiction severity, suggesting that state-resolved features can track clinical dimensions. This convergence of group differences and severity associations on a specific state suggests that the disorder's core signature lies in the internal organization of a particular integrative configuration rather than a global reorganization of the resting brain.

The paper concludes that dynamic, state-resolved designs are necessary to detect these network alterations, which are likely diluted by static, time-averaged analyses. However, the authors maintain a modest stance, noting that the cross-sectional, single-site design limits causal inference and generalizability, and that replication in larger, independent, and longitudinal samples is required before these state-specific patterns can be interpreted mechanistically or used as clinical biomarkers.

More like this