22.8 Conclusion

Adaptive instruction has a long history (Reiser, 1987). However, systematic effort aimed at developing adaptive instructional systems was not made until the early 1900s. The effort for developing adaptive instructional systems has taken different approaches: macro-adaptive, ATI, and micro-adaptive. Macro-adaptive systems have been developed to provide more individualized instruction on the basis of the student's basic teaming needs and abilities determined prior to instruction. The ATI approach is to adapt instructional methods, procedures, or strategies to the student's specific aptitude information. Micro-adaptive systems have been developed to diagnose the student's teaming needs and provide optimal instructional treatments during the instructional transaction process.

Some macro-adaptive instructional systems seemed to be positioned as an alternative educational system because of their demonstrated effectiveness. However, most of the macrosystems were discontinued without much success because of the difficulty associated with the development and implementation of the systems, including curriculum development, teacher training, resource limitation, and organizational resistance. Numerous studies have been conducted to investigate ATI methods and strategies because of ATI's theoretical appealing and practical application possibilities. However, the results are not consistent and have provided little implications for developing an adaptive instructional system.

Using computer technology, a number of different micro-adaptive instructional systems have been developed. However, their applications have been mostly in laboratory environments because of the limitation of their functional capability to handle the complex transaction processes involved in the teaming of various types of tasks by many different students.

Another contribution to the limited success of adaptive instructional systems can be attributed to the unverified theoretical assumptions that were used for the development of the systems. Particularly, ATI, including the achievement and treatment interactions, has been used as the theoretical bases for many ATI studies. However, the variability of ATI research findings suggests that the theoretical assumptions used in ATI research may not be valid, and the development of a complete taxonomy of all likely aptitudes and instructional variables may not be possible. Even if it is possible to develop such a taxonomy, its instructional value will be limited because teaming will be influenced by many variables, including aptitudes. Also, the instructional value of aptitude variables measured prior to instruction becomes less important as the instruction progresses. In the meantime, the student's on-task performance (i.e., response to the given problem or task) becomes more important for diagnosing the student's teaming needs (see Fig. 22-1) because on-task performance is the integrated reflection of many verifiable and unverifiable variables involved in the teaming.

Therefore, I propose an on-task performance and treatment interaction approach. In this approach, response-sensitive methods will be used as the primary strategy. Many studies (e.g., Atkinson, 1974; Park & Tennyson, 1980, 1986) demonstrated the effects of response-sensitive strategies. However, the application of the response-sensitive strategy has been limited to simple tasks such as vocabulary acquisition and concept teaming because of the technical limitations of handling the complex interactions involved in teaming more sophisticated tasks such as problem solving. However, ITSs developed in the last 2 decades have demonstrated that technical methods and tools are now available for the development of more sophisticated response-sensitive systems. Unfortunately, this technical development has not significantly contributed to an intellectual breakthrough in the field of learning and instruction. Thus, no principles or systematic guidelines for developing questions and explanations necessary in the response-sensitive strategy have been developed. In this chapter, I reviewed several studies that provide some valuable suggestions for the development of response-sensitive strategies, including asking diagnostic questions and providing explanations (Collins & Stevens, 1983; Brown & Palincsar, 1989; Leinhardt, 1983). Further research for asking diagnostic questions and providing explanations is needed for the development of response-sensitive adaptive systems.

Since response-sensitive diagnostic and prescriptive processes should be developed on the basis of many different types of information available in the system, I propose to use a complete model of adaptive instructional systems described by Park et al. (1987). This model consists of input, transactions, and output stages, and components directly required to implement the response-sensitive strategy are in the transaction stage of instruction.

To develop an adaptive instructional system using this model will require a multidisciplinary approach because it will need expertise from different domain areas such as learning psychology, cognitive science or knowledge engineering, and instructional technology (Park & Seidel, 1989). However, with the current technology and our knowledge of learning and instruction, the development of a complete adaptive instructional system like the one presented in Figure 22-3 may not be possible in the immediate future. It is expected that cognitive scientists will further improve the capabilities of current Al technology such as natural language dialogues and inferencing processes for capturing the human reasoning and cognitive process. In the meantime, the continuous accumulation of research findings in learning and instruction will make a significant contribution to instructional researchers' and developers' efforts for developing more powerful adaptive instructional systems.