Case-based reasoning and learning

Case-based reasoning is a computational model that uses prior experiences to understand and solve new problems. The foundation of the CBR system is laid on Schank's arguments on the role of reminding (1982), which coordinates past events with current events to enable generalization and prediction. The assumption is that when a case is indexed to a place where another experience is already indexed, a reminding happens and the potential for generalization exists (Kolodner & Kolodner, 1987). Different from expert systems, which store past experience as generalized rules and objects, CBR systems store past experiences as individual problem solving episodes (Kolodner, 1992; Maher, Balachandran, & Zhang, 1995). CBR systems also attempt to generate solutions to a new problem based on the use of these experiences from previously solved problems. The underlying principle of such a system is that people solve new problems by remembering similar experiences about similar situations.

According to Schank (1990), human memory is story-based. What people know is stored in memory as stories. People are reminded of past experiences by current ones, and we use those past experiences as a guide to help us process new experiences. Kolodner (1993) defines a case as "a contextualized piece of knowledge representing an experience that teach a lesson fundamental to achieving the goal of the reasoner." Thus, in the CBR paradigm, problem solving is viewed as a process of remembering a specific problem-solving episode, adapting the solution to fit the current situation, and storing the adapted solution in the memory. With this problem solving process, the solved problems with the adapted solutions can be indexed as new cases into learners' memory for future use. It is exactly the accumulation of the cases that demonstrates the acquisition of the expertise. Put simply, learning occurs when people process new experiences in light of old ones (Shank, 1990). Similarly, Kolodner et al. (1996) stated that learning means extending one's knowledge by incorporating new experiences into memory, by re-indexing old experiences to make them more accessible, and by abstracting out generalizations from experiences.
Moreover, studies on the nature of expertise all show that experts differ from novices in the amount of their knowledge, the organization and accessibility of the knowledge, and the methods used to apply the knowledge (Chase & Simon, 1973; Chi, Feltovich, & Glaser, 1981; Gobbo & Chi, 1986). ). According to the knowledge representation in the CBR systems, the critical factor that sets experts apart from novices is the ability of experts to deal effectively with new situations by recalling and reusing the relevant experience. Therefore, it is thought to be valuable to use the CBR system to teach novices, who do not have much problem solving experience, by presenting stories of others in the problem-solving context.

Analogical reasoning in case-based reasoning
In order to assist learning from prior problem-solving episodes, the CBR systems should be able to execute a basic process: retrieving, adapting and storing. Montazemi & Gupta (1997), although focusing on the system's decision making about how to retrieve a similar case, elaborate the process between adapting and storing, i.e., evaluating the adapted solution, and predicting success or failure of the solution. Incorporated with Leake's concept of analogy (1996), a process model of the CBR system is proposed (See Figure 1: A process model of the CBR system).

The proposed CBR process emphasizes the performance of analogical reasoning and the feedback of evaluation in order for a case-based reasoner to learn its lessons while adding a new experiential episode of success or failure to its memory. As Leake (1996) pointed out, case-base reasoning can be viewed as fundamentally analogical. Didierjean and Cauzinille-Marmeche (1998) discussed that there are two processesunderlying reasoning by analogy: one is to use abstract knowledge; the other is to use case-based reasoning. Previous psychological experiments show that human reasoners do use both processes of reasoning by analogy simultaneously (Goldin-Meadow et al., 1993; Didierjean & Canzinille-Marmeche, 1998).

In case-based reasoning, the degree of relevance of the retrieved case to the new situation is a crucial element to the reasoning process. Thus, the research on CBR is more concerned with the issue of indexing to form correspondence between the new experience episode and a previous one. The index is important because an index for a case allows a reminding strategy to recognize situations in which the case is relevant. Kolodner (1993) made a very clear statement about the importance of such indexing. She defined a case as "a contextualized piece of knowledge representing an experience that teaches a lesson fundamental to achieving the goal of the reasoner" (p. 15). The key element for learners to remember in applying the previous cases is how much information an index is able to capture about those cases so that the situations to which the cases apply can be identified. Thus, in the CBR approach, solving a problem consists of searching out the nearest case and then adapting it without mediation of a more abstract knowledge structure (Hammond, 1990). From this perspective, this case based analogical reasoning contrasts with rule-based learning. The process of case-based reasoning includes:

Teaching with cases
Case-based reasoning (CBR), which originated from the field of artificial intelligence, promotes learning through cases. Support for teaching with cases comes from a number of areas of research in cognitive science. Situated cognition, for example, emphasizes the importance of the context in which knowledge is used. Thus, learning should be situated in the culture of the practitioners (Brown, Collins, & Dugid, 1989). From this perspective, cases, consisting of setting, actors, goals, and a sequence of events (Kolodner & Guzia, 2000), are useful in instruction because they can involve learners in analyzing specific problems presented in a realistic scenario.

CBR derives its theoretical support from memory organization and reminding in cognitive science (Shank, 1982; 1990). CBR uses prior experiences in the process of problem solving. CBR, as a learning model, argues that the acquisition of expertise is to accumulate experiences with a succession of real cases and to properly index these experiences for later retrieval (Koschmann et al, 1997). The power of the reasoning activities through the access to old cases provides a potential instructional practice in problem solving. In fact, Kolodner, Hmelo, and Narayanan (1996) proposed to incorporate CBR with PBL to enhance learning. They suggested that a library of cases that cover an adequate variety of the problem solving experiences from others should be presented at the moment when the cases are needed. It is expected that the case library, providing the experiences that human learners lack, will be able to augment learners' memory and thus enhance learners' problem solving and reasoning skills. Moreover, the theoretical support for the learning strength in CBR is derived from the ability of reasoning the old problem solving episodes to navigate the new ones. A problem solver or a case-based reasoner demonstrates reasoning ability in the process of interpreting and adapting the solution through analyzing the similarities and differences

Kolodner (1993) listed the advantages of case-based reasoning (p.25-27):


Brown, J. S., Colins, A., & Dugid, P. (1989). Situated cognition and the culture of learning. Educational Researcher, 18 (1), 32-42.

Chase, W. G., & Simon, H. A. (1973). Perception in chess. Cognitive Psychology, 4, 55-81.

Chi, M. T. H., Feltovich, P. J. & Glaser, R. (1981). Categorization and representation of physics problems by experts and novices. Cognitive Science, 5, 121-152.

Chi, M. T. H., Lewis, M. W., Reimann, P., & Glaser, R. (1989). Self-explanations: How students study and use examples in learning to solve problems. Cognitive Science, 13, 145-182.

Didierjean, A., & Cauzinille-Marmeche, E. (1998). Reasoning by analogy: Is it schema-mediated or case-based? European Journal of Psychology of Education. 8 (3), 385-398.

Domeshek, E., & Kolodner, J. (1993). Using the point of large cases. Artificial Intelligence for Engineering Design, Analysis and Manufacturing, 7(2), 87-96.

Edelson, D. C. (1993). Learning from stories: Indexing and Reminding in a Socratic case-based teaching system for elementary school biology (Doctoral dissertation, Northwestern University, 1993). UMI's Dissertation Abstracts.

Gick, M. L., & Holyoak, K. J. (1980). Analogical problem solving. Cognitive Psychology, 12, 306-355.

Gobbo, C., & Chi, M. (1986). How knowledge is structured and used by expert and novice children. Cognitive Development, 1, 221-237.

Goldin-Meadow, S., Nusbaum, H., Garber, P., & Chruch, R. B. (1993). Transition in learning: Evidence for simultaneously activated strategies. Journal of Experimental Psychology: Human Perception and Performance, 19(1), 92-107.

Hammond, K. J. (1990). Case-based planning: A framework for planning from experience. Cognitive Science, 14, 385-443.

Kolodner, J. L. (1992). An introduction to case-based reasoning. Artificial Intelligence Review, 6(3), 3-34.

Kolodner, J. L. & Guzdial, M. (2000). Theory and practice of case-based learning aids. In D. H. Jonassen, & S. M. Land (Eds.), Theoretical foundations of learning environments (pp. 215-242). Mahwah, NJ: Lawrence Erlbaum Associates.

Kolodner, J. L., & Kolodner, R. M. (1987). Using experience in clinical problem solving: Introduction and framework. IEEE Transactions on Systems, Man and Cybernetics, 17(3), 420-431.

Koshmann, T., Collins, A. Klein, G, Holyoak, K., & Kolodner, J. (1997). The role of cases in learning. Proceedings of the 19th Annual Conference of the Cognitive Science Society. p.979.

Leake, D. B. (1996). CBR in context: The present and future. Retrieved April 10, 2000 from the Wide Web:

Maher, M. L., Balachandran, M. B., & Zhang. D. M. (1995). Case-based reasoning in design. Mahwah, NJ: Lawrence Erlbaum Associates.

Montazemi, A. R. & Gupta, K. M. (1997). A framework for retrieval in case-based reasoning systems, Annals of Operations Research, 72, 51-73.

Shank, R. C. (1982). Dynamic memory: A theory of learning in people and computers. Cambridge: Cambridge University Press.

Shank, R. C. (1990). Tell ma a story: Narrative and intelligence. Evanston, IL: Northwestern University Process.