Closing the Cognitive Gap between Humans and Interactive Narrative Agents Using Shared Mental Models
Rania Hodhod, Brian Magerko · 2016
This paper proposes a new formal approach for negotiating shared mental models between humans and computational improvisational agents (improv agents) based on our sociocognitive studies of human improvisers. Negotiation of shared mental models serves as a core mechanism for improv agents to co-create stories with each other and with human interactors. The model aims to narrow the gap between human and machine intelligence by providing AI agents that, in the presence of incomplete knowledge about an improv scene, can use procedural representations not only to understand human parties but also to negotiate their mental models with them. The described approach allows flexible modeling of ambiguous, non-Boolean knowledge through the use of fuzzy logic and situation calculus that allows reasoning under uncertainty in a dynamic improvisational setting.