[Paper Review] Understanding opinions. A cognitive and formal account

Francesca Giardini, Walter Quattrociocchi+1 more|arXiv (Cornell University)|Jun 21, 2011

Opinion Dynamics and Social Influence23 references3 citations

TL;DR

This paper proposes a cognitively grounded model of opinions as mental representations with three core features: truth value, perceived truth value, and confidence level, formalized using time-varying graphs to capture dynamic opinion change. It distinguishes evaluative and factual opinions, models social influence through relational connected components, and offers a formal framework that integrates cognitive processes with social dynamics, overcoming limitations of prior reductionist models.

ABSTRACT

The study of opinions, their formation and change, is one of the defining topics addressed by social psychology, but in recent years other disciplines, as computer science and complexity, have addressed this challenge. Despite the flourishing of different models and theories in both fields, several key questions still remain unanswered. The aim of this paper is to challenge the current theories on opinion by putting forward a cognitively grounded model where opinions are described as specific mental representations whose main properties are put forward. A comparison with reputation will be also presented.

Motivation & Objective

• To address the lack of a clear cognitive and formal definition of opinions in social psychology and computational models.
• To distinguish evaluative from factual opinions based on cognitive features such as truth value and confidence.
• To develop a formal model using time-varying graphs that captures opinion dynamics at both individual and social levels.
• To integrate micro-level cognitive processes with macro-level social influence, avoiding reductionism.
• To provide a foundation for simulating opinion formation, revision, and diffusion in heterogeneous agent systems.

Proposed method

• Defining opinions as mental representations with three components: objective truth value (T_o), perceived truth value (T_s), and confidence (d_c).
• Using time-varying graphs (TVG) to model evolving social networks where opinion dynamics unfold over time.
• Introducing the concept of a 'relational-connected component' induced by external stimuli, representing the subset of mental nodes activated during opinion comparison.
• Formalizing opinion change as a function of the confidence and truth value alignment within connected components of an agent’s epistemic network.
• Modeling social influence through bounded confidence, where agents interact only with others within a tolerance interval [x−ε, x+ε].
• Applying the framework to simulate opinion diffusion and revision in multi-agent systems, with parameters derived from cognitive plausibility.

Experimental results

Research questions

• RQ1 How can opinions be formally defined as mental representations with distinct cognitive features?
• RQ2 What are the cognitive mechanisms that determine resistance to opinion change?
• RQ3 How do social interactions and network structure influence opinion dynamics in a cognitively plausible way?
• RQ4 In what way do evaluative and factual opinions differ in their formation and transformation processes?
• RQ5 Can time-varying graphs effectively model the dynamic interplay between individual cognition and social influence in opinion systems?

Key findings

• Opinions are best understood as mental representations with three defining features: objective truth value, perceived truth value, and confidence level, which together determine resistance to change.
• The model distinguishes evaluative opinions (value-laden) from factual opinions (truth-based), each governed by different cognitive dynamics.
• Opinion change is driven by the activation of relational-connected components in an agent’s epistemic network, where alignment of truth values and confidence levels determines persuasion likelihood.
• The use of time-varying graphs enables modeling of dynamic social networks and evolving opinion interactions, overcoming limitations of static metrics and formalisms.
• Bounded confidence is formalized as a tolerance interval around an opinion, where interaction and influence occur only within this range, reflecting cognitive constraints on social acceptance.
• The framework provides a foundation for simulating opinion evolution in multi-agent systems, enabling exploration of opinion persistence, diffusion, and transformation under cognitive and social constraints.