Mobile phone short video use negatively impacts attention functions: an EEG study

The study investigates whether and how tendencies toward mobile phone short-form video addiction affect attentional functions and their neural underpinnings. Grounded in the tripartite attention network model (alerting, orienting, executive control) assessed by the Attention Network Test (ANT), the authors hypothesized that higher short-form video addiction tendencies would be associated with impaired attention, particularly executive control, reflected in EEG theta-band activity. The work addresses a gap in understanding the neural mechanisms linking short-form video overuse to attentional deficits observed behaviorally in prior studies.

Prior research on substance use and behavioral addictions shows impaired inhibitory control and attentional biases (Murphy and Garavan, 2011; Wilcockson and Pothos, 2015; Dong et al., 2017; Wang et al., 2017; Bowling et al., 2020; Wilcockson et al., 2021). Excessive smartphone/short-form video use has been linked to social withdrawal and possible attention problems and health risks in adolescents and others (Tateno et al., 2019; Liu et al., 2021; Chen et al., 2022). Some studies report more attention deficits among heavy short-form video users and impaired attentional concentration during interference (Qi et al., 2022; Zhou et al., 2022). The ANT assesses alerting, orienting, and executive control (Fan et al., 2002, 2005; Petersen and Posner, 2012), with neural substrates spanning frontoparietal, salience, and prefrontal/anterior cingulate circuits (Posner, 2012; Proskovec et al., 2018; Bowling et al., 2020; Markett et al., 2022). Theta-band oscillations, especially frontal/midfrontal theta, are implicated in attention and cognitive control, including conflict monitoring/resolution (Aftanas and Golocheikine, 2001; Botvinick et al., 2004; Fan et al., 2007; Cavanagh and Frank, 2014; Wei and Zhou, 2020). Despite behavioral links between short-form video use and attention deficits, the neural mechanisms, particularly EEG correlates during attention tasks, remained unclear, motivating the present EEG-ANT study focusing on theta activity.

Design and participants: Cross-sectional EEG study using the ANT. Forty-eight healthy adults (13 males, 35 females; age 18–33 years, M = 21.80, SD = 3.62) who used mobile short-form video apps were recruited via WeChat advertisements. Inclusion criteria: 18–65 years; no neurological/psychiatric diagnoses; no psychotropic drug/alcohol use in past month; no severe adverse reactions to flicker, auditory, or electromagnetic stimuli; no prior similar EEG study participation. Ethical approval was obtained from Zhejiang University; informed consent was collected.

Questionnaires: Participants completed an online battery including MPSVATQ (short-form video addiction tendency), IAT (Internet Addiction Test), SCS (Self-Control Scale), ACS (Attention Control Scale), BIS-II (impulsivity), CPSS (perceived stress), Depression and Anxiety Brief Symptom Survey, FFMQ (mindfulness), and MWQ (mind-wandering).

Experimental protocol: After questionnaires, two 3-minute eyes-open resting-state EEG sessions were recorded, one before and one after ANT. The ANT lasted ~10 minutes.

ANT task: 192 trials: 4 cue types (no, center, double, spatial; 48 trials each) × 3 target types (neutral, congruent, incongruent; 16 each under each cue). Fixation 400–1600 ms (random), cue 100 ms (or none), then target (row of five arrows/lines) above/below fixation; target remained until response or 1700 ms max; total trial duration fixed at 4000 ms via variable post-response interval. Participants identified the central arrow direction (left = F key; right = J key). Practice: 12 trials per session; ≥85% accuracy required to proceed (all passed; up to 3 practice sessions for 3 participants). Speed and accuracy emphasized; no feedback.

EEG acquisition: 64-channel cap (Neuroscan Quick-cap; 10–20 system), Neuroscan Synamps2 amplifier, 1000 Hz sampling. Vertical and horizontal EOG recorded. Impedances <10 kΩ. Shielded lab.

Preprocessing: EEGLAB used. Task EEG: remove redundant electrodes (CB1, CB2) and EOG channels; downsample to 250 Hz; re-reference to M1/M2; bandpass 0.1–35 Hz; notch 49–51 Hz; epoch −200 to 1300 ms relative to cue onset; ICA to remove ocular/cardiac artifacts. Resting-state EEG: same preprocessing steps except no epoching.

Time–frequency analyses: For ANT task epochs, Short-Time Fourier Transform (STFT) with 0.5 s window to obtain time–frequency representations, focusing on theta (4–8 Hz). Time windows of interest (TWI): alerting 650–800 ms post-cue; orienting 600–800 ms post-cue; executive control 600–900 ms post-cue. Regions of interest (ROIs): for alerting/orienting—parietal (P5, P3, P1, Pz, P2, P4, P6), parieto-occipital (PO5, PO3, POz, PO4, PO6), occipital (O1, Oz, O2); for executive control—frontal (F5, F3, F1, Fz, F2, F4, F6), frontocentral (FC3, FC1, FCz, FC2, FC4), central (C5, C3, Cz, C2, C4), parietal (as above), parieto-occipital (as above). Mean amplitudes computed across selected electrodes and trials; log-mode baseline correction used. Resting-state EEG: FFT; absolute power in dB; theta band (4–8 Hz) in same ROIs as task analysis.

Behavioral metrics and statistics: Extreme RT/ACC values >3 SD from mean were replaced by the corresponding average per standard ANT practice. RMANOVA 4 (cue) × 3 (target) on RT and ACC. ANT network scores: Alerting = RT(no cue) − RT(double cue); Orienting = RT(center cue) − RT(spatial cue); Inhibitory effect = RT(incongruent) − RT(congruent). Correlations between MPSVATQ and ANT network scores.

EEG metrics and statistics: Neural oscillation indexes mirroring behavioral network scores computed using theta power instead of RT: Alerting = power(no cue) − power(double cue); Orienting = power(center cue) − power(spatial cue); Inhibitory = power(incongruent) − power(congruent). Additionally, correlations between MPSVATQ and theta power difference under incongruent minus neutral targets were examined, focusing on ROIs. Covariates: gender, age, anxiety, depression. Correlations between SCS and neural indices were also tested. Resting-state theta power correlations with MPSVATQ were assessed. Significance threshold p < 0.05 with Bonferroni correction where applicable. Software: SPSS 22.0 and MATLAB 2013b. Note: EEG data from 3 participants excluded due to excessive movement artifacts; EEG analyses included N = 45.

• Sample and questionnaires: N = 48 (13 males, 35 females; M age = 21.80, SD = 3.62). MPSVATQ correlated positively with IAT (r = 0.390, p < 0.01) and negatively with SCS (r = −0.320, p < 0.05) and ACS (r = −0.310, p < 0.05).
• ANT behavioral performance (RMANOVA): RT showed significant main effects of cue (F(3,141) = 70.20, p < 0.001, partial η² = 0.60) and target (F(2,94) = 134.40, p < 0.001, partial η² = 0.74), and a significant cue × target interaction (F(6,282) = 4.59, p < 0.001, partial η² = 0.09). No-cue had the longest RT; spatial cue the shortest; incongruent RT > congruent. ACC showed significant main effects of cue (F(3,141) = 10.27, p < 0.001, partial η² = 0.18) and target (F(2,94) = 48.47, p < 0.001, partial η² = 0.51), and a significant interaction (F(6,282) = 2.60, p = 0.018, partial η² = 0.05); center-cue ACC < other cues; spatial cue highest; incongruent ACC < congruent.
• No significant correlations between MPSVATQ and behavioral ANT network efficiencies (alerting, orienting, inhibitory effects; all p > 0.05).
• EEG during ANT (N = 45): No significant correlations between MPSVATQ and theta-based neural indices of alerting, orienting, or inhibitory effects defined by standard contrasts (all p > 0.05).
• Executive-control related theta effect: Significant negative correlation between MPSVATQ and frontal theta power difference for incongruent minus neutral targets (r = −0.395, p = 0.007) within 600–900 ms post-cue at frontal electrodes (F5, F3, F1, Fz, F2, F4, F6). Other regions showed non-significant trends (middle frontal r = −0.280, p = 0.063; central r = −0.229, p = 0.130; parietal r = −0.014, p = 0.926; parieto-occipital r = −0.118, p = 0.439). The frontal correlation remained significant after controlling for gender, age, anxiety, and depression.
• Self-control and EEG: No significant correlation between SCS and the frontal theta incongruent-minus-neutral difference (p > 0.05).
• Resting-state EEG: Theta power in prefrontal region from two resting sessions showed no significant correlation with MPSVATQ (p > 0.05) and did not differ between sessions (t(43) = 1.42, p = 0.888).

Despite robust ANT effects on RT and ACC across cue and target conditions, short-form video addiction tendency (MPSVATQ) did not relate to behavioral measures of attentional network efficiency, possibly due to task simplicity and low variance. Crucially, higher addiction tendency was associated with reduced frontal theta modulation during executive control demands (incongruent versus neutral), indicating compromised neural processes supporting conflict monitoring and resolution in prefrontal circuitry. This aligns with literature showing neural differences in addicted populations without overt behavioral deficits and underscores the role of frontal theta in cognitive control. The effect persisted after adjusting for demographic and affective covariates and was task-specific, as resting-state theta showed no relationship with MPSVATQ. The negative association between MPSVATQ and self-control (SCS) further supports a link between short-form video addiction tendencies and diminished self-regulatory capacity. Together, the findings suggest that short video overuse may alter neural dynamics of executive control even when behavior appears intact, highlighting targets for intervention (e.g., enhancing frontal theta-related control processes).

The study demonstrates that greater mobile phone short-form video addiction tendency is associated with reduced frontal theta power during executive control demands in the ANT, suggesting impaired executive control at the neural level. Additionally, higher addiction tendency correlates with lower self-control. Although behavioral ANT indices did not correlate with addiction tendency, the task-dependent EEG findings point to altered prefrontal control mechanisms. These insights support development of interventions (e.g., mindfulness-based training) to bolster self-control and potentially remediate neural deficits linked to short video addiction.

• Cross-sectional design limits causal inference; longitudinal studies are recommended.
• Modest, young adult sample (N = 48; EEG N = 45) with gender imbalance limits generalizability; broader, more diverse, and gender-balanced samples, including children and older adults, are needed.
• The study focused on a limited set of short-form video behaviors; expanding to different platforms and categories (e.g., YouTube Shorts) would enhance generalizability.
• EEG provides limited spatial resolution; multimodal imaging (e.g., fMRI) could better localize neural alterations.
• Additional behavioral and psychosocial factors (e.g., daily usage time, social and psychological variables) should be incorporated in future models.
• Intervention studies are needed to test strategies aimed at mitigating effects of short video addiction on attention.