Modelling the impact of beliefs and communication on attitude - - PowerPoint PPT Presentation

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Modelling the impact of beliefs and communication on attitude dynamics: a cognitive agent-based approach

Kei-Léo Brousmiche1,3 Jean-Daniel Kant2 Nicolas Sabouret2 Stephane Fournier3 François Prenot-Guinard3

1LIP6 - Université Pierre & Marie Curie, Paris, France 2LIMSI-CNRS - Université Paris-Sud, Orsay, France 3Airbus Defense & Space, Elancourt, France

Journées des thèses, Val de Reuil, 2014

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 1

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Context: Stabilization Operations

Needs for new systems Population-centric training systems Perception-attitude-behaviour dynamics toward Forces Social simulation Attitude dynamics Agent-based modelling

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 2

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Objectives

Model of attitude dynamics based on Psychological theories Representation of beliefs Model of communication Belief exchange Social network Inter-ethnic conflicts Model social groups

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 3

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Classical Attitude Models in Social Simulation

Simple models e.g binary, discrete or real values [Nowak et al., 1990] ✪ Do not consider the construction mechanism of the attitude [Urbig and Malitz, 2007] Sum of the evaluations of the object’s features ✪ Bounded-confidence model ⇒ attitude-beliefs connections are lost ✪ Agents shouldn’t have unllimited memory

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 4

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

An Attitude Models in Psychology

Attitude as object-evaluation associations [Fazio, 2007] ✦ Links between attitude and beliefs ✦ Limited memory and varying accessibilies

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 5

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Communication

What to communicate? Attitude itself [Xia et al., 2010, Castellano et al., 2009] Or part of [Urbig and Malitz, 2007, Thiriot and Kant, 2007]

• However:

no psychological theory describes the impact of attitude exchange conversational narratives represents up to 40% of daily communication [Eggins and Slade, 1997]

✇ Belief exchange To whom? Small-world [Milgram, 1967] Social groups

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 6

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

General Principle

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 7

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Agent Cognition

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 8

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Communication and Belief Revision

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 9 1

Credibility: direct > population > Force

2

Belief revision when individual i receives a(sender) [Thiriot and Kant, 2008]:

– action unknown, accept a(sender) – know compatible, reinforce a(i) – know incompatible, probability to accept a(sender)

3

Heuristic of information relay: based on Credibility and Recency

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Benefit Evaluation

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 10

Agents evaluate actions based on its payoff and their attitudes toward the beneficiaries. eval(a, i) = OWAip∈impactList(a,i)(payoff(ip) × att(i, subject(ip))) Evaluation model inspired from [Ajzen and Fishbein, 2005] OWA: Ordered Weighted Average operator [Yager, 1988]

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Accessibility of a Belief

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 11

Peak-End Theory [Kahneman et al., 2003] to select “most accessible” beliefs acc(a, i) = α × credibility(a, i) + β × unexpectedness(a, i) + γ × impactAmplitude(a, i) + δ × recency(a, i)

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Attitude construction

att(i, f) = tanh

• ρ.OWAa(i)∈aList(i) (eval(a, i) × acc(a, i))
• kei-leo.brousmiche@lip6.fr

LIP6, Airbus D&S, LIMSI 12

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Basic Behavior

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 13

100 agents ; 3 social groups ; 1 Force ; 3 actions

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Discordant Information Broadcast

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 14

INTRODUCTION RELATED WORKS MODEL EVALUATION CONCLUSION

Conclusion

We proposed an attitude dynamics based on socio-psychological theories a construction of attitude that relies on beliefs with beliefs revision through communication a generic model Futur works more socio-psychological theories less parameters introduction of emotion real case study

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 15

References I

Ajzen, I. and Fishbein, M. (2005). The influence of attitudes on behavior. In Albarrac, D., Johnson, B. T., and Zanna, M. P ., editors, The handbook of attitudes, pages 173–221. Lawrence Erlbaum Associates Publishers, Mahwah, NJ, US. Castellano, C., Fortunato, S., and Loreto, V. (2009). Statistical physics of social dynamics. Reviews of modern physics, 81(2):591. Eggins, S. and Slade, D. (1997). Analysing casual conversation. Equinox Publishing Ltd. Fazio, R. H. (2007). Attitudes as object-evaluation associations of varying strength. Social Cognition, 25(5):603. Kahneman, D., Kahneman, D., and Tversky, A. (2003). Experienced utility and objective happiness: A moment-based approach. The psychology of economic decisions, 1:187–208.

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 16

References II

Milgram, S. (1967). The small world problem. Psychology today, 2(1):60–67. Nowak, A., Szamrej, J., and Latane, B. (1990). From private attitude to public opinion: A dynamic theory of social impact. Psychological Review, 97(3):362. Thiriot, S. and Kant, J.-D. (2007). Représenter les croyances par des réseaux associatifs pour simuler la diffusion d’innovations. ESSA, 7:2. Thiriot, S. and Kant, J.-D. (2008). Reproducing stylized facts of word-of-mouth with a naturalistic multi-agent model. In Second World Congress on Social Simulation, volume 240. Urbig, D. and Malitz, R. (2007). Drifting to more extreme but balanced attitudes: Multidimensional attitudes and selective exposure. ESSA, Toulouse.

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 17

References III

Xia, H., Wang, H., and Xuan, Z. (2010). Opinion dynamics: Disciplinary origins, recent developments, and a view on future trends. Yager, R. R. (1988). On ordered weighted averaging aggregation operators in multicriteria decisionmaking. Systems, Man and Cybernetics, IEEE Transactions on, 18(1):183–190.

kei-leo.brousmiche@lip6.fr LIP6, Airbus D&S, LIMSI 18