The next speaker in this session at the 2026 Association of Internet Researchers conference in Mexico City is Frederik Weber Henriksen, whose interest is in cross-cutting engagement and polarisation. This is often connected to people sorting into homogeneous information environments (or ‘echo chambers’), yet cross-cutting exposure is actually very common in everyday experience, and such exposure can itself also be a source of polarisation.
Passive and active cross-cutting engagement and polarisation turn out to be more complex matters than simple ‘echo chamber’ and ‘filter bubble’ narratives tend to suggest, not least also because such cross-cutting exposure can also occur in the context of critical, counter-attitudinal content forwarding. Simple network analyses cannot capture this sufficiently.
Exposure is not engagement; critical sharing enables the circulation of content well beyond its intended recipients; in response, many users also actively curate their personal feeds. User responses to the content they encounter can include influence as well as unfollowing; even when seeing disagreeable content we sometimes move as slightly towards such views; and how a message is framed in network communication contexts also matters a great deal.
This project, then, explores these questions by operationalising exposure, circulation, interpretation, and following and unfollowing actions in the context of oppositional content sharing, simulated via agent-based modelling. This is tracked via the extremity and speed of polarisation, and an analysis of partisan network structures.
The project simulates an experimental setup of 500 users in a directed preferential-attachment network; these may post, and read the last 10 items, for for each content encounter have a choice of possible actions. Several parameters can be adjusted for each simulated user.
In this simulation, if oppositional sharing of content is allowed for these agents, this turns out to accelerate the process of polarisation; further, differences in how such oppositional shares are perceived also affect the outcome: if the oppositional share is perceived as a mixed message, this polarises less than if the oppositional sharer’s comment and the shared content are perceived as separate messages.
These simulation results now need to be validated against real-life contexts, but provide a useful starting-point for further studies that can also vary the platform feed mechanisms and affordances and underlying network structures.
