Personalization and attentional bias in problematic short-video use

Short-video platforms such as TikTok rely on personalized recommendation algorithms that curate highly tailored content streams. Although personalization is widely assumed to be associated with problematic use, the cognitive mechanisms linking personalization to problematic engagement remain unclear. The present study examined attentional selectivity in response to personalized short videos and how this selectivity varies with levels of problematic use.

We conducted an eye-tracking experiment using the dot-probe task in which individuals with higher and lower levels of problematic use viewed personalized screenshots drawn from their own recommendation feeds and general non-personalized screenshots. To eliminate stimulus confounds, each participant's personalized stimuli served as another participant's non-personalized stimuli.

A significant interaction between user group and stimulus type was revealed. Individuals with lower levels of problematic use displayed selective attentional bias toward personalized content, whereas individuals with higher levels of problematic use showed comparable bias toward personalized and non-personalized content. Eye-tracking indices mirrored this pattern, indicating lower selectivity in gaze allocation among individuals with higher versus lower levels of problematic use. To further quantify this pattern, we computed attentional selectivity indices reflecting the difference in bias toward personalized versus non-personalized stimuli. Higher problematic-use scores significantly correlated with lower attentional selectivity.

These findings suggest that individuals with higher problematic short-video use exhibit less differentiated attentional responses to personalized versus non-personalized short-video cues, with attentional bias generalizing beyond personally tailored cues. This pattern highlights attentional selectivity as a potential cognitive process linking personalization and problematic short-video engagement.