33. 11 An Instructional Theory Of Learner Control?

Can a comprehensive, integrative, deductive, prescriptive, and testable theory of learner control be developed? I suspect not, not if we demand of such a theory (any more than any other educational theory) that it be falsifiable in the Popper (1968) sense. An alternative question, however, is whether we can still develop instructional prescriptions for the use of learner control which are at least pragmatic and are grounded in some reasonable psychological and educational principles. I suspect we probably can. In fact, work to develop such prescriptions can be found in many of the existing learner-control reviews.

Steinberg (1984), for example, lists a range of these events that might be offered within a learner-controlled lesson, together with some conditions that might mitigate their success: which topics to study and in what order, number of exercises to practice and their level of difficulty, presentation of review or supplementary materials, or the option not to answer questions. Other activities, too, could be made optional: amount or kind of feedback to see following practice questions, whether to exit the instruction, mode of presentation (e.g., verbal or graphic), and even the option of whether to allow further learner control at all.

Laurillard (1987) presents another assortment of computer-based learning strategies of which learners might judiciously be given control. One category of these strategies, control of content sequence, includes provisions for the student to skip forward or backward a chosen amount or to retrace a route through the material, and options to control when to view such features as content indexes or content maps. [A rather remarkable early example of learner control of content sequence in computer-based instruction comes from Grubb (1968). He describes a system whereby the student, with the aid of a light pen and a content map on the screen, is able to point and jump to any subtopic in the lesson. This approach presages the current "hypertext" environments in which students proceed through instruction in a nonlinear "browsing" fashion.]

Another category presented by Laurillard (1987) is called control of learning activities, and includes options for the student to see examples, do exercises, receive information, consult a glossary, ask for more explanation, and take a quiz. Most of her list of learner-controlled activities is included in Steinberg's (1984) list, but Laurillard's seems more complete and grounded in educational-psychological theory.

Milheirn and Martin (1991) discuss when and how to prescribe three types of variables for which students might be granted control: control of pacing, that is, the speed of presentation of instructional materials; and control of content, permitting students to skip over certain instructional units. They suggest that these categories, in addition to control of sequence (similar to Lautillard's control of content sequence), represent the most germane sets of instructional variables affecting the success or failure of learner-controlled CBI.

Recently others, most notably Chung and Reigeluth (1992), have worked to synthesize an empirically based and pragmatic listing of instructional prescriptions that link a variety of learner-control strategies (over content, sequence, pace, display/strategy, internal processing, and use of advisor systems) to instructional conditions (learner characteristics, learning objectives or domains, and instructional systems) and broad outcomes (learning achievement, transfer, and retention, time efficiency, cost efficiency, and attitudes toward learning and instruction). These authors recognize the multidimensional nature of learner control, and provide a helpful set of do's and don'ts for deciding when to employ which learner-controlled instructional events. Their recommendations are many, and only a few are presented here as illustrations of their approach.

For example, they recommend that students should be offered control of content when they have significant previous knowledge in the content area, because presentation of already known material could be irrelevant and interesting. If students are already prepared with some content knowledge, they can more effectively manage their own content. Additionally, content control might be given when the learning objectives are of a higher-order type, as opposed to factual information.

Similarly, they list conditions that allow for learner control of sequence, such as when the instructional program is quite lengthy (sequence control can help maintain learner motivation and interest), or when they are familiar with a topic. Likewise, students should not be given control over sequence when learning objectives have a clear prerequisite order, or when it would be impractical to break up and resequence existing materials.

In a similar vein, they provide many conditions for using learner control of specific instructional elements such as pacing, displaying information or using instructional strategies, internal processing (including some metacognition strategies), and use of advisory systems. To justify many of these prescriptions, Chung and Reigeluth (1992) cite empirical studies from the learner-control literature; many others, however, are derived from current instructional theories (e.g., Gagn6, 1985; Merrill, 1983) and need empirical validation in their own right. Given the overall structure they give to their instructional prescriptions, essentially a series of "if-then" conditions, it would be interesting to see developed a type of computer-based decision tree or expert system based on the "mix and match" combinations of instructional strategies, outcomes, and conditions presented in their paper. Such a prescriptive system of learner control might then be validated with research across a variety of instructional systems and contexts, and would provide some tests of the generalizability of their recommendations.

All of the categorization schemes providing advice on if, how, and when to use learner control in CBI overlap to a large degree and differ primarily in perspective or orientation. And all provide useful information for designers attempting to decide whether and how to include learner-controlled events in their instructional designs. None, however, is comprehensive or definitive.