The Great Deceivers: Virtual Agents and Believable Lies

Abstract

This paper proposes a model giving Theory of Mind (ToM) capabilities to artificial agents to allow them to carry out deceptive behaviours. It describes a model supporting an N-level Theory of Mind and reports a study to assess whether equip- ping agents with a two-level ToM results in them being perceived as more socially intelligent than agents with a single-level ToM. A deception game being developed for intercultural training of children, used for this study, is described. Finally, we report results from this study consistent with the hypothesis that a two-level Theory of Mind better supports agents in deceptive behaviour.


Back to Table of Contents