Deciding to Remind during Collaborative Problem Solving: Empirical Evidence for Agent Strategies

No. 1: Agents, AI in Art and Entertainment, Knowledge Representation, and Learning

Volume

Proceedings of the AAAI Conference on Artificial Intelligence, 13

Interaction

Previous work suggests that reminding a conversational partner of mutually known information depends on the conversants’ attentional state, their resource limits and the resource demands of the task. In this paper, we propose and evaluate several models of how an agent decides whether or not to communicate a reminder. We elaborate on previous findings by exploring how attentional state and resource bounds are incorporated into the decision making process so that reminders aid the performance of agents during collaborative problem solving. We test two main hypotheses using a multi-agent problem solving simulation testbed: (1) an agent decides to present salient knowledge only when it reduces overall problem solving effort (2) an agent can use its own attentional state as a model of the attentional state of its partner when assessing the effort trade-offs of communicating a reminder. Our results support both hypotheses, suggesting that the models we propose should be further tested for multi-agent communication in problem solving situations.