Learner control for adaptive learning: The importance of learners' perceptions

The research studies presented in this dissertation focus on computer-based instruction and how learner control (LC) can assist in enhancing the learning process and learning outcomes. Fast changes in computational technologies have led to the development of highly adaptive computer-based learning environments. Adaptive learning environments, such as intelligent tutoring systems, have been shown to be effective for learning, although there are some inherent drawbacks to such environments. Due to the unidirectional adaptation from system to learner, one of the core principles of instructional design –involvement of learners– might be undermined. De Corte, Verschaffel and Lowyck (1994) for instance, stated that highly adaptive learning environments can lack opportunities for active and constructive learner involvement. In line with this, Elen (2000) expressed the concern that learners might then not be able to develop their self-regulation skills and are at risk to become dependent of the prestructuralized instruction that is offered in adaptive systems. Consequently, learners are not invited to take responsibility of their learning process (Hannafin & Land, 1997). Already in 1975, Merrill stated that the goal of instructional design should be to make learners system-independent rather than system-dependent. One way to deal with this is providing learners with LC over their learning and offer adaptable learning environments. LC as instructional strategy has received broad attention and was incorporated in several instructional design theories such as the 4C/ID model (van Merriënboer, Clark, & de Croock, 2002), the elaboration theory of instruction (Reigeluth & Stein, 1983), and the self-determination theory (Deci & Ryan, 2000), all emphasizing the importance of learners’ engagement during learning. Providing the learners with control and choice is assumed to enhance intrinsic motivation, to stimulate the development of self-regulation strategies and hence to enhance effective and efficient learning. The notion of LC in computer-based instruction has many facets and LC can be defined along two dimensions. As Scheiter and Gerjets (2007) described, there are several types of LC (e.g., control over pacing, content, presentation). Also LC can vary in the degree of control that is available (e.g., full learner control, shared control, advice). Although the overall conclusion is that providing the learners with some control is more effective than no control at all (Goforth, 1994), empirical research on LC in computer-based instruction reveals mixed results. LC as instructional strategy is not equally effective for all learners and there is the risk that already good learners will benefit more from LC than less skilled learners (Hannafin, 1984). Recently, Corbalan, Kester, and van Merriënboer (2009) defined three threats to effective LC: (1) LC can only be effective when learners have the necessary skills to deal with LC (e.g., prior knowledge and self-regulation skills); (2) LC can only be effective if it is perceived as such by learners (e.g.,learners interpret instructions which in turn can influence their learning behaviour); and (3) LC demands free cognitive resources to process and may thus impose additional cognitive load in learners during learning. This dissertation focuses on the three prerequisites and takes into account learners’ skills, perceptions and experienced cognitive load in trying to unravel the mechanisms behind effective LC. Hence, the main research question addressed in this dissertation is how the instructional design of LC is related with learners’ perceptions of LC and their learning outcomes. To investigate this research question, several subquestions are addressed: (1) to what extent does instruction affect learners’ perceptions of LC and adaptive learning; (2) whether LC induces cognitive load, and more specifically, what type of cognitive load; (3) to what extent are features of the learning materials related with learners’ perception of LC (4) whether providing learners with advice on the use of control can enhance learning outcomes; and (5) can LC be redefined towards LC on a metacognitive level through the use of open learner models (OLMs)?The first part of this dissertation provides a review (chapter 2) of research that includes learner characteristics such as prior knowledge, motivation, learning behaviour, etc. to develop adaptive learning environments. Next to this, chapter 3 presents a research study wherein it is investigated whether adaptive systems are effective because of the adaptivity as such, or also because of learners perceiving this adaptivity. Part two of this dissertation then focuses on adaptable, learner-controlled learning. In this part, four research studies are presented that investigate whether LC imposes cognitive load to the learners (chapter 4); whether instruction of LC affects learning (chapter 5); whether LC effectiveness might be related to features of the learning materials, such as task characteristics (chapter 6); and whether advice on LC (in terms of tool use) can assist in more effective LC (chapter 7).In part three, this dissertation concludes with redefining LC as instructional strategy. First, the added value of open learner models is discussed (chapter 8). OLMs provide learners with information on the rules behind the adaptivity of the system and let learners interact with the system as to adjust the way adaptive instruction is offered. As such, OLMs provide learners with control on a metacognitive level, rather than on the level of content and presentation selection. To conclude (chapter 9), new research lines are discussed in which both usability design research and instructional design research can assist in investigating the mechanisms behind effective LC in computer-based instruction.nrpages: 268status: publishe