Experimental Investigation Of An Agent Commitment Strategy

In dynamic environments, optimal deliberation in the decision-theoretic sense is impossible. Instead, it is sometimes necessary to trade potential decision quality for decision timeliness. One approach to achieving this trade-o is to endow intelligent agents with meta-level strategies that provide them guidance about when to reason| and what to reason about|and when to act instead. In this paper, we describe our investigations of a particular meta-level reasoning strategy, ltering, in which an agent commits to the goals it has already adopted, and then tends to lter from consideration new options that would con ict with the successful completion of existing goals [Bratman et al. 1988]. To investigate the utility of ltering, we conducted a series of experiments using the Tileworld testbed [Pollack and Ringuette 1990]. Previous experiments [Kinny and George 1991] provided preliminary evidence of the feasibility of ltering; our results generalize and re ne those earlier claims to show the types of conditions under which ltering leads to improved performance, and the types of conditions that require restrictions on ltering.