CASE-BASED LEARNING - PREDICTIVE FEATURES IN INDEXING

Interest in psychological experimentation from the Artificial Intellig ence community often takes the form of rigorous post-hoc evaluation of completed computer models. Through an example of our own collaborativ e research, we advocate a different view of how psychology and Al may be mutually relevant, and propose an integrated approach to the study of learning in humans and machines. We begin with the problem of learn ing appropriate indices for storing and retrieving information from me mory. From a planning task perspective, the most useful indices may be those that predict potential problems and access relevant plans in me mory, improving the planner's ability to predict and avoid planning fa ilures. This ''predictive features'' hypothesis is then supported as a psychological claim, with results showing that such features offer an advantage in terms of the selectivity of reminding because they more distinctively characterize planning situations where differing plans a re appropriate. We present a specific case-based model of plan executi on, RUNNER, along with its indices for recognizing when to select part icular plans-appropriateness conditions-and how these predictive indic es serve to enhance learning. We then discuss how this predictive feat ures claim as implemented in the RUNNER model is then tested in a seco nd set of psychological studies. The results show that learning approp riateness conditions results in greater success in recognizing when a past plan is in fact relevant in current processing, and produces more reliable recall of the related information. This form of collaboratio n has resulted in a unique integration of computational and empirical efforts to create a model of case-based learning.