22.7 A model of adaptive instructional systems

In the above section, I emphasized the importance of on-task performance or a response-sensitive approach in the development of adaptive instructional systems. However, a complete adaptive system should have the capability to update continuously every component in the instructional system based on the student's on-task performance and the interactions between the student and system. However, almost all adaptive instructional systems, including ITSs, have been developed with emphasis on a few specific aspects or functions of instruction. Therefore, we present a conceptual model for developing a complete adaptive instructional system (Fig. 22-3). This model is adopted from the work of Seidel and his associates (Seidel, 1971), with consideration of recent developments in learning and instructional psychology and computer technology (Park, Perez & Seidel, 1987).

This model does not provide specific procedures or technical guidelines for developing an adaptive system. However, we think that the cybernetic metasystem approach used in the model is generalizable as a guide for developing a more effective and efficient control process required in adaptive instructional systems. The model illustrates what components an adaptive system should have and how those components should be interrelated in an instructional process. Also, the model shows what specific self-improving or updating capabilities the system may need to have.

As Figure 22-3 shows, this model divides the instructional process into three stages: input, transactions, and output. The input stage basically consists of the analysis of the student's entry characteristics. The student's entry characteristics include not only his or her within-lesson history (e.g., response history) but also prelesson characteristics. The prelesson characteristics may include information about the student's aptitudes and other variables influencing his or her teaming. As discussed earlier, the aptitude variables measured prior to instruction will be useful for the beginning stage of instruction but will become less important as the student's on-task performance history is accumulated. Thus, the within-lesson history should be continuously updated using the evaluation information of the performance (i.e., output measures).

The transaction stage consists of the interactions between the student and system. In the beginning stage of the instruction, the system will select problems and explanations to present based on the student's entry characteristics, mainly the premeasured aptitudes. Then, the system will evaluate the student's responses (or any other student input such as questions or comments) to the given problem or task. The response evaluation provides information for diagnosing the student's specific teaming needs and for assessing overall performance level on the task. The teaming needs will be inferred according to diagnostic rules in the system. Finally, the system will select new display presentations and questions for the student according to the tutorial rules. The tutorial rules should be developed in consideration of different teaming and instructional theories (e.g., see Snelbecker, 1974; Reigeluth, 1983), research findings (e.g., see Gallangher, 1994; Weinstein & Mayer, 1986), expert heuristics (Jonassen, 1988), and response-sensitive strategies discussed in the earlier section of this chapter.

The output stage mainly consists of performance evaluation. The performance evaluation may include not only the student's overall achievement level on a given task and specific performance on the subtasks but also the analysis of complete teaming behaviors related to the task and subtasks. According to the performance evaluation and analysis, the instructional components will be modified or updated. The instructional components to be updated may include contents in the knowledge base (including questions and explanations), instructional strategies, diagnostic and tutorial rules, the lesson structure, and entry characteristics. If the system does not have the capability to modify or update some of the instructional components automatically, the human monitor may be required to perform that task.