On short video platforms, individuals can not only passively browse videos but also actively engage in behaviors such as liking and commenting on videos. However, the mechanisms by how active and passive usage of short videos affect subjective well-being remains unclear. Thus, this study aims to explore these mechanisms through path analysis and network analysis. We employed the Short Video Usage Questionnaire, Subjective Well-Being Scale, and Interaction Anxiety Scale to survey 1086 participants. The results indicated (1) the active usage positively predicted life satisfaction and positive affect, while passive usage failed to predict any of them. (2) Social anxiety mediated the relationship between short video usage and subjective well-being. Active usage enhanced life satisfaction and positive affect by alleviating social anxiety, while reducing negative affect by decreasing social anxiety. At the same time, passive usage reduced life satisfaction and positive affect by exacerbating social anxiety while increasing negative affect by intensifying social anxiety. (3) Network analysis revealed that “live interaction” behavior was a core activity for alleviating social anxiety and enhancing subjective well-being. The findings highlight the significant role of short video usage in subjective well-being and provide empirical evidence to guide users in the rational and positive usage of short video platforms.

Visual search can be guided and facilitated by a category-specific attentional template (CAT) when searching for a target embedded into a set of categories. While previous research has proven that CATs can be pre-activated, it has caused confusion due to inconsistent task difficulty and unbalanced target-defined dimensions. To avoid the unbalanced dimensions and investigate the effects of category frameworks on the time course of the pre-activation of CAT, we defined the targets as a color-defined category based on prototype (warm, cool), semantic (garden, ocean), and strategy (red–blue, yellow–green). Moreover, to maintain the consistency of task difficulty, we employed the identification task to all frameworks of CAT within the RSPP paradigm and tracked the time course of CAT pre-activation by measuring N2pcs elicited by the probes sequentially occurring prior to the target. ERP and MVPA results found that the semantic-based CAT (involving in experience and learning) was pre-activated first, approximately 1200 ms before the search display, followed by the prototype-based CAT (involving in perceptual similarity) at about 900 ms, and finally the strategy-based CAT (involving in learning) at around 600 ms before the search display. By balancing the target dimensions and maintaining the difficulty of tasks, the findings provide further evidence for differences in the activation patterns of various constitutive CATs.