Does online sharing drive attention to extreme viewpoints? [Registered Report Stage 1]

Political polarization is often attributed to selective exposure, yet the mechanism by which online environments amplify extreme positions remains unclear. Attention is a precondition for downstream beliefs and behaviors, but existing theories of how it is allocated across political content have rarely been compared directly. In this Registered Report, we specify and test six theoretical accounts of attentional selection: selective exposure, the confrontation effect, decision difficulty, political acrophily, audience-response utility, and task-dependent attention, alongside a null account in which none of the theoretical predictors explain attention. Participants recruited online and stratified by political affiliation view 88 political posts, presented one at a time in a microblogging format, while webcam-based eye tracking measures visual attention. The ideological position of each post is defined as the median placement assigned by independent pretest raters. In the following condition, participants decide for each post whether they would follow the author; in the sharing condition, whether they would share the post on their own social media. After completing the main task, participants rate each post on personal agreement and perceived social-circle agreement. The theoretical accounts are represented by a preregistered set of mixed-effects models and compared using Bayes factors. The design tests whether selecting a source for oneself, compared with sharing a post with an audience, influences attention and, if so, which mechanism best describes it.