Opinions

In human activities, opinions are everywhere. Politics, prices, religion, even science, all are activities are influenced by them (whether they should be or not is another question entirely).

Part of understanding those problems requires that we have good theoretical models of how we change our minds, by observing others and also by making any general observations about the world. Opinion dynamics is the sub-area of Sociophysics where we study some of those problems.

Big questions

Consensus

When can we expect a group to reach consensus?

Is always agreement a good thing?

Extremism

How extremism can spread in a society?

How can extremism emerge from moderate opinions?

How to counter dangerous extreme points of view?

Opinions and Truth

Social influence can hinder pursuit for truth?

Opinions and fake news

My models, my answers

Emergence of Extremism

CODA model

By introducing a probabilistic strength to opinions about binary choices, Continuous Opinions and Discrete Actions showed how extremists can appear from simple local reinforcement rules.

How extreme can depend on network structure, external agents, mobility, and other factors.

Supported by: FAPESP Grant 2008/00383-9

Types of agents

The literature in OD has introduced different types of agents to model specific circumstances. Contrarians were studied in the context of CODA.

CODA also was helpful at understanding the emergence of inflexibles.

Supported by: FAPESP Grants 2008/00383-9 and 2009/08186-0

Expanding on the theme

A Framework

Extending CODA allowed the proposal of a general framework for opinion models.

Traditional models can be obtained in limit cases if CODA agents consider their own influence on their neighbors.

A CODA-like version for continuous opinions shows the same behavior as Bounded Confidence models.

Supported by : FAPESP Grants 2008/00383-9 and 2009/08186-0

Applications

CODA was used as basis for understanding interesting problems such as:

Diffusion of innovations.

Acceptance of scientific theories (see also)

Supported by : FAPESP Grants 2008/00383-9 and 2009/08186-0

Communication and Mental Models

What agents expect to conclude can be more important than what is communicated

Estimating trust based on ideological agreement can make a difference

Supported by : FAPESP Grant 2014/00551-0

Bayesian inspired but not CODA

Continuous opinion observed as continuous value on a range.

We show, using perceptrons, that many issues can lead to opinion dispersal.

Supported by: FAPESP Grant 2008/00383-9


Trust and network evolution

Trust can destroy tendency to consensus even in a fully connected network and despite its much slower evolution.

Opinions and networks evolve together. Models should reflect that.

Discrete Multiple Options with strength

When there are many independent choices, agent still can have extreme opinions.

Defining Extremism

Current theoretical models do not agree on how to define extremist opinions. Here is my take on that.

Interested?