Social Norm Emergence in Virtual Agent Societies

  • Authors:
  • Bastin Tony Savarimuthu;Maryam Purvis;Martin Purvis;Stephen Cranefield

  • Affiliations:
  • Department of Information Science, University of Otago, Dunedin, Dunedin, New Zealand;Department of Information Science, University of Otago, Dunedin, Dunedin, New Zealand;Department of Information Science, University of Otago, Dunedin, Dunedin, New Zealand;Department of Information Science, University of Otago, Dunedin, Dunedin, New Zealand

  • Venue:
  • Declarative Agent Languages and Technologies VI
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

The advent of virtual environments such as SecondLife call for a distributed approach for norm emergence and spreading. In open virtual environments, monitoring various interacting agents (avatars), using a centralized authority might be computationally expensive. The number of possible states and actions of an agent could be huge. An approach for sustaining order and smoother functioning of these environments can be facilitated through norms. Agents can generate norms based on interactions. In particular, those social norms that incur certain cost to an individual agent but benefit the whole society are more interesting than those benefit both the agent and the society. The problem is that the selfish agents might not be willing to share the norm adherence cost. In this work, we experiment with notion proposed by Axelrod that social norms are best at preventing small defections where the cost of enforcement is low. We also study how common knowledge can be used to facilitate the overall benefit of the society. We believe our work can be used to facilitate norm emergence in virtual online societies.