[Paper Review] Opinion dynamics model based on cognitive biases

Paweł Sobkowicz|arXiv (Cornell University)|Mar 4, 2017

Opinion Dynamics and Social Influence94 references3 citations

TL;DR

This paper introduces an agent-based model of opinion dynamics that incorporates cognitive biases—particularly confirmation bias and politically motivated reasoning—using Bayesian updating to simulate how individuals filter information. The model demonstrates that when biases are strong, consensus on objective truth fails to form, even with overwhelming evidence, while malleable biases (like politically motivated reasoning) can be shifted through normative re-framing, enabling convergence.

ABSTRACT

We present an introduction to a novel model of an individual and group opinion dynamics, taking into account different ways in which different sources of information are filtered due to cognitive biases. The agent based model, using Bayesian updating of the individual belief distribution, is based on the recent psychology work by Dan Kahan. Open nature of the model allows to study the effects of both static and time-dependent biases and information processing filters. In particular, the paper compares the effects of two important psychological mechanisms: the confirmation bias and the politically motivated reasoning. Depending on the effectiveness of the information filtering (agent bias), the agents confronted with an objective information source may either reach a consensus based on the truth, or remain divided despite the evidence. In general, the model might provide an understanding into the increasingly polarized modern societies, especially as it allows mixing of different types of filters: psychological, social, and algorithmic.

Motivation & Objective

• To develop a realistic model of opinion dynamics that accounts for cognitive biases in information processing.
• To investigate how different types of cognitive biases—especially confirmation bias and politically motivated reasoning—affect consensus formation.
• To explore whether and how opinion polarization can be mitigated by altering group norms or information framing.
• To extend traditional agent-based models by embedding psychological mechanisms like Bayesian belief updating with biased likelihood functions.
• To enable simulation of complex social dynamics involving psychological, social, and algorithmic information filters.

Proposed method

• Agents update their belief distributions using Bayesian updating, where prior beliefs and likelihood ratios are influenced by cognitive biases.
• The model incorporates two key biases: confirmation bias (prior-dependent filtering) and politically motivated reasoning (PMR, where likelihood assessment is driven by identity-congruent beliefs).
• Information sources are modeled as objective distributions S(θ), which agents filter based on their internal bias parameters.
• The model allows for both static and time-dependent bias functions, enabling analysis of long-term opinion stability and transient responses.
• Agents can be embedded in social networks, allowing for bidirectional communication and asymmetric information processing based on differing filter types.
• The framework supports dynamic redefinition of in-group norms, which can alter PMR filters and trigger shifts in group identity or opinion alignment.

Experimental results

Research questions

• RQ1 How do confirmation bias and politically motivated reasoning differ in their impact on consensus formation under objective information?
• RQ2 Can politically motivated reasoning be modified through changes in perceived in-group norms, leading to convergence on objective truth?
• RQ3 What role do dynamic group norms play in shifting opinion dynamics and reducing polarization?
• RQ4 How do multiple conflicting information sources affect opinion evolution when filtered through different cognitive biases?
• RQ5 In what ways can external manipulation of information filters increase or reduce societal polarization?

Key findings

• When agents are governed by confirmation bias, consensus on objective truth fails to emerge, even with repeated exposure to the same objective information.
• In contrast, when politically motivated reasoning is the dominant filter, opinion convergence on the truth is possible if the perceived in-group norms are redefined to align with objective evidence.
• The model shows that PMR is more malleable than confirmation bias because it is driven by identity-congruence rather than fixed priors, making it responsive to normative changes.
• Re-framing information to align with in-group values can alter the likelihood function in Bayesian updating, thereby shifting posterior beliefs without changing priors.
• The model supports the possibility of reducing polarization not by changing beliefs directly, but by altering the social context that shapes information filtering.
• Dynamic redefinition of in-group membership can trigger shifts in opinion dynamics, transforming from opinion change to changes in group identity and cohesion.