The Effects of Rapid Assessments and Adaptive Restudy Prompts in Multimedia Learning.

Introduction Educational technology provides us with learning environments which adapt learning paths to the individual's needs. Students learn by problem solving via a number of well-established systems. Their corresponding attempts can be used to adapt hints and learning tasks to their individual needs (e.g., Cognitive Tutor; http://www.carnegielearning.com; Koedinger & Corbett, 2006). Other environments provide expository instruction and present multimedia contents to learners. In these cases, explicit probes or diagnostic tasks must be interspersed for adaptation purposes (e.g., SmartBook; http://www.engadget.com/2013/01/08/mcgraw-hill-smartbook). An efficient means of such diagnosis is the use of rapid assessment procedures as developed by Kalyuga and colleagues within the framework of cognitive load theory (e.g., Kalyuga, 2008; Kalyuga & Sweller, 2005). Rapid assessment tasks can be used to detect learners' knowledge gaps and to adapt further learning paths to these deficits. We conducted two experiments on the effects of rapid assessment tasks and two types of adaptive restudy prompts in multimedia learning. Rapid assessments are tasks that should be fulfilled quickly and that are interspersed throughout a learning environment. The first experiment addressed the extent to which rapid assessment tasks can be regarded as a non-reactive diagnostic method, meaning that they do not change or influence what they measure (i.e., knowledge states). Note that in the first experiment, rapid assessment was not used to adapt instruction; we tested just its potential reactivity. In the second experiment, rapid assessment was used to diagnose knowledge gaps and adapt instruction. Specifically, we tested the effects of two types of restudy prompts triggered by wrong responses to rapid assessment tasks. Adaptive learning systems If you confront different learners with a learning environment, they are sure to differ in their learning outcomes (e.g., Ackerman & Lohman, 2006). These differences can be attributed to the learners' varying prerequisites and the fact that a one-size-fits-all environment cannot be optimal for different learning prerequisites. One remedy is to use adaptive learning environments (e.g., Shute & Zapato-Rivera, 2008; Vandewaetere, Desmet, & Clarebout, 2011). Adaptation can refer to different sizes of grain in this context. At its coarsest, macro-adaptation refers to assigning different learning environments to different learners (Park & Lee, 2003). When the grain is fine, micro-adaptation refers to adapting instructional events during learning to a learner's cognitive or affective states. We focused on micro-adaptation in our experiments. Adaptive systems can react to different learner characteristics such as (prior) knowledge states, working memory capacity, cognitive styles, motivation, or emotional states (see, e.g., Vandewaetere et al., 2011). Knowledge-related variables have most frequently been used to adapt instruction (e.g., prior knowledge, knowledge states, or identified knowledge gaps). This emphasis on knowledge-related factors is not surprising, given their conceptual affinity with the knowledge-related learning goals of most adaptive systems. Furthermore, knowledge prerequisites are the most important factor for further learning both positively and negatively. Correct prior knowledge is usually the most important factor facilitating further learning (e.g., Dochy, de Rijdt, & Dyck, 2002; Kalyuga, 2012). Incorrect knowledge (e.g., misconceptions or misunderstanding) is usually the most substantial barrier for further learning (Ambrose & Lovett, 2014). In addition, research on aptitude-treatment interactions and on the expertise-reversal effect (i.e., sensible instructional features for novices lose their effectiveness with more knowledgeable learners) has clearly shown that different instructional procedures are best used for learners with different knowledge states (for an overview see Lee & Kalyuga, 2014). …