Case-based learning: Predictive features in indexing

Abstract

Abstract. Interest in psychological experimentation from the Artificial Intelligence community often takes the form of rigorous post-hoc evaluation of completed computer models. Through an example of our own collaborative research, we advocate a different view of how psychology and AI may be mutually relevant, and propose an integrated approach to the study of learning in humans and machines. We begin with the problem of learning appropriate indices for storing and retrieving information from memory. From a planning task perspective, the most useful indices may be those that predict potential problems and access relevant plans in memory, improving the planner's ability to predict and avoid planning failures. 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 are appropriate.We present a specific case-based model of plan execution, RUNNER, along with its indices for recognizing when to select particular plans--appropriateness conditions--and how these predictive indices serve to enhance learning. We then discuss how this predictive features claim as implemented in the RUNNER model is then tested in a second set of psychological studies. The results show that learning appropriateness 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 collaboration has resulted in a unique integration of computational and empirical efforts to create a model of case-based learning.

Finding the right case: the role of predictive features in memory for aviation accidents

“…These results strongly support the view that some aspects of narrative accounts inherently make better indices than others. In particular, the results are consistent with the 'predictive features' model (Seifert et al, 1994) that suggests that the most useful index in a narrative is the complication or problem that leads to the resulting actions and outcomes. In this study, however, although there was a very clear-cut difference between the utility of the complication feature and the setting and outcome features, the complication feature was not significantly different from the decision feature.…”

“…Experience-related superiority in recall is generally attributed to the development of conceptual organization (Ericsson & Kintsch, 1995;Herbert & Burt, 2001). The importance of the complication feature in recall lends some support to Seifert et al's (1994) predictive features theory. Contrary to expectations however, both a nonpredictive feature (setting information) and a predictive feature (complication information) were effective for expert participants (pilots) and nonexpert participants alike.…”

“…In order to be retrieved, cases must have been tagged or indexed in some way that predicts their relevance to the current event. Seifert et al (1994) proposed that a functional system would index cases according to the features that were most predictive of future actions. They define this as 'when conditions and constraints are known, but a decision has not yet been made ' (p. 42).…”

“…It was hypothesized that pilots would recall a greater proportion of information from the cases than nonpilots. Following the 'predictive features' model of Seifert et al (1994), it was hypothesized that pilots would recall a greater percentage of information from the cases when provided with the complication feature than any of the other three features. It was expected that differences between the feature types would be less pronounced for nonpilots.…”

A theoretical framework for goal-based choice and for prescriptive analysis

“…is an oft-heard refrain from those who have to decide what action to take. The retrieval of "similar cases" from memory is a focus on the study of expertise (e.g., Seifert et al 2002). Creativity, or "thinking outside the box," is recognized when the set of options offered or retrieved initially from memory is rejected and a rather different option is proposed.…”