A Dialogical Logic-based Simulation Architecture for Social Agents - - PowerPoint PPT Presentation

A Dialogical Logic-based Simulation Architecture for Social Agents

and the Emergence of Extremist Behaviour Piter Dykstra1,2, Corinna Elsenbroich3, Wander Jager1, Gerard Renardel de Lavalette1, Rineke Verbrugge1

1Groningen University,

Groningen ,The Netherlands

2Hanzehogeschool Groningen,

Groningen ,The Netherlands

3University of Surrey,

Guildford, United Kingdom

September 10, 2009

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks

1 Introduction 2 Groups & Extreme Opinions

The ”Henhouse”-model The Demo

3 Argumentation

Dialogical Logic Model with Dialogues

4 Concluding Remarks

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks

Central Question

What causes the emergence of extreme opinions? Goal Agent-based Simulation with Reasoning Agents

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks

Central Question

What causes the emergence of extreme opinions? Goal Agent-based Simulation with Reasoning Agents

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks The ”Henhouse”-model

Actions

Announcements

Loudness Evidence Importance

Adopt opinion from other agents (assimilation) Move elsewhere (segregation)

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks The ”Henhouse”-model

The Topic Space

Simulation of Common Knowledge

Accumulation of evidence and importance Forgetting

Formation of groups of agents

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks The ”Henhouse”-model

The Topic Space

Simulation of Common Knowledge

Accumulation of evidence and importance Forgetting

Formation of groups of agents Reputation Status is determined by similarity with environment. Goal: Maximize RS

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks The Demo

RationalAgent program in an initial state

(A) (B) (C) Agents Evidence Importance

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks The Demo

RationalAgent with social knowledge

(A) (B) (C) Agents Evidence Importance

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks The Demo

Results I

Emergence of extreme opinions Decrease of importance of opinions (and communication) Formation of (different kinds of) groups. No reasoning or argumentation involved

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks Dialogical Logic

Dialogues in Rational Agents

Proponent - Opponent Initial Announcement Attacks Defences Alternating Moves Winning Pay - Reward

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks Dialogical Logic

Dialogues

Proponent - Opponent Initial Announcement Attacks Defences Alternating Moves Winning Pay - Reward Reputation Status is determined by winning and losing dialogues.

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks Dialogical Logic

Example of a Dialogue

A: If we do not have enough funding and refugees create additional costs, then they should leave this country. B: Why do you say that? A: Well, prove me wrong! If you can I give you 1RS but if I am right you have to pay me 0.50RS. B: Ok, deal. I really disagree with your conclusion. . . . A: Do you agree then that we do not have enough funding and that refugees create additional costs. B: Yes of course, that can hardly be denied. A: Right. . . – people, do you believe refugees should leave this country?

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks Dialogical Logic

2-D Logic

Evidence: E(ϕ) =

P(ϕ) P(ϕ)+R(ϕ)

Importance: I(ϕ) = P(ϕ)+R(ϕ)

2

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks Dialogical Logic

2-D Logic

Evidence: E(ϕ) =

P(ϕ) P(ϕ)+R(ϕ)

Importance: I(ϕ) = P(ϕ)+R(ϕ)

2

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks Model with Dialogues

RationalAgent with dialogues

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks Model with Dialogues

Results II

No extremization of opinions No decrease of importance One agent wins all the RS-points and determines the ruling opinion Communication stops

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks

Logic in Rational Agents

Epistemic Logic for Higher Order Reasoning Pitfall: the perfect reasoner Dialogical Non-Monotonic Multi-Valued Paraconsistent

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture

Introduction Groups & Extreme Opinions Argumentation Concluding Remarks

The Competition

Social Network Analysis (Sun and Breiger) Belief-Desire-Intention (Flache & Macy) Game Theory (van Benthem) Constructuralism (Carley)

Piter Dykstra et al. A Dialogical Logic-based Simulation Architecture