Coherent belief and opinion propagation produces more echo chambers: corrected version

Hiro Kataoka, J{é}r{ô}me Euzenat, Koji HasebearXiv 2026 · arXiv

TL;DR

This paper explores how the interplay between our opinions (how we feel) and our beliefs (what we think is true) accelerates the formation of echo chambers in social networks. By simulating agents that strive for internal coherence between their views and their justifications, the researchers demonstrate that adding "logical" beliefs to simple opinion models leads to more fragmented and segregated digital communities.

What Problem Does This Paper Address?

Most research on social media polarization uses "opinion dynamics" models, which treat views as simple numbers (e.g., a scale from 0 to 1). While useful, these models ignore a crucial human element: we don't just hold opinions; we hold beliefs that justify them. For instance, you might support a new transit line (opinion) because you believe it will reduce traffic (belief).

The authors address the gap between symbolic logic (beliefs) and numeric values (opinions). They investigate whether the psychological drive to keep these two aligned—a concept called coherence—makes echo chambers more likely to form. They define an echo chamber through three specific lenses:

  1. Homogeneity: Everyone in the group shares very similar views.
  2. Segregation: The group mostly interacts with itself and ignores outsiders.
  3. Reinforcement: Views become more similar and entrenched over time.

Methods

The authors developed a multi-agent simulation that integrates four distinct processes:

  • Opinion Dynamics (OD): Based on the "bounded confidence" model, where agents move their numeric opinions closer to neighbors who are already somewhat similar.
  • Belief Revision (BR): Using propositional logic, agents merge their logical formulas with those of their neighbors to reach a consistent set of beliefs.
  • Coherence Restoration: This is the paper's novel contribution. They introduced two operations: Opinion Formation (OF), where an agent updates its opinion to match its new beliefs, and Belief Alignment (BA), where an agent searches for new beliefs to justify a changed opinion.
  • Cultural Values: To bridge the gap between logic and numbers, they use a "value map" that assigns numeric scores to different logical states of the world.

Agents also perform network rewiring, a process where they "unfollow" people with different views and "follow" those who are more like-minded, as illustrated in the network evolution below:

Visualization of a sample network before and after rewiring.

Datasets

The study relies on synthetic data generated through multi-agent simulations rather than real-world social media scrapes.

  • Scale: 100 agents with 400 initial connections.
  • Duration: 5,000 iterations per simulation.
  • Variables: The researchers varied "tolerance" thresholds (how different someone can be before you stop listening to them) for both opinions ($\epsilon$) and beliefs ($\delta$).

Key Insights and Ideas

  • The Double Constraint: When agents require both similar opinions and similar beliefs to interact, the "concordance" bar is much higher. This naturally leads to more fragmentation.
  • Coherence as a Wedge: The drive for internal consistency (keeping beliefs and opinions aligned) acts as a powerful engine for segregation. If an agent changes an opinion, they must find new beliefs to justify it, which might move them further away from a neighbor who hasn't made that same logical leap.
  • Smaller, More Numerous Chambers: Interestingly, while enforcing coherence increases the number of belief-based echo chambers, it often results in smaller groups. The social fabric doesn't just tear in half; it shatters into many tiny, highly consistent shards.

Key Results

The simulations confirmed that introducing beliefs significantly changes network topology:

  • Confirmed Hypotheses: Both opinions and beliefs independently form echo chambers. Connecting them (the "both" condition) increases the total number of chambers compared to looking at opinions alone.
  • The Impact of Coherence: Maintaining coherence (the S3 setting) led to the highest number of belief echo chambers (an average of 2.85 compared to 1.91 when beliefs evolved in isolation).
  • Tolerance Sensitivity: As shown in the plots below, as agents become more "tolerant" (higher $\epsilon$ or $\delta$), the number of echo chambers drops, but the "coherence" model (S3) consistently produces more chambers than simpler models at low tolerance levels.
Effect of tolerance thresholds on the number of echo chambers.

The final state of these networks often shows clear clusters where color (representing views) is perfectly uniform within a group but distinct from others:

Final network states across different experimental settings.

Limitations and Future Work

  • Simplified Logic: The model uses a very basic logic system (only 3 atoms/propositions). Real-world belief systems involve thousands of interconnected ideas.
  • Static Values: The "Cultural Values" map used to link beliefs to opinions is the same for all agents and never changes. In reality, people's underlying values are diverse and can shift over time.
  • Stability vs. Perturbation: The authors suggest future work should test how "robust" these echo chambers are. If you introduce a "fact-checker" agent or a major external event, do these coherent clusters break apart or double down?

Summary updated

Abstract

The occurrence of echo chambers is a well-known phenomenon in social networks. Echo chambers are groups of agents satisfying three properties: homogeneity, i.e., sharing the same opinion, segregation, i.e., listening primarily to agents of the same group, and reinforcement, i.e., increasing opinion similarity over time. Opinion dynamics models have been used to explain how echo chambers occur through agents updating their opinions. However, social network users also exchange beliefs supporting their opinions. Here, we aim at identifying the effect of propagating beliefs together with opinions on the formation of echo chambers. For that purpose, we complement an existing opinion dynamics model with classical belief merge operations. Beliefs and opinions are also maintained coherent guided by values that agents share. We introduce measures, applying to either beliefs and opinions, to identify echo chambers based on the three properties. This model is used to simulate opinion and belief propagation with varying initial conditions and agent tolerance for different opinions/beliefs. Results confirm the occurrence of echo chambers observed in previous opinion dynamics studies in a different setting and with a stricter measure. They also demonstrate the occurrence of echo chambers with belief propagation. Finally, they show that connecting opinions to beliefs and maintaining their coherence increases the number of belief echo chambers, but that they involve a reduced number of agents.
BibTeX
@misc{kataoka2026coherent,
  author = {Hiro Kataoka and J\{é\}r\{ô\}me Euzenat and Koji Hasebe},
  title = {{Coherent belief and opinion propagation produces more echo chambers: corrected version}},
  year = {2026},
  eprint = {2611.25879},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2611.25879},
}
Figure