Enhancing Simulation-Based Learning through Active External Integration of Representations

Enhancing Simulation-Based Learning through Active External Integration of Representations Daniel Bodemer (d.bodemer@iwm-kmrc.de) Knowledge Media Research Center, Konrad-Adenauer-Str. 40 72072 Tuebingen, Germany oriented way and therefore interact with the simulations rather randomly (e.g., de Jong & van Joolingen, 1998; Schauble, Glaser, Raghavan, & Reiner, 1991). Additional problems may be caused by the dynamic visualization of the simulated concepts. On the one hand the externalization of dynamic processes may prevent learners from performing cognitive processes relevant to learning on their own (e.g., Schnotz et al., 1999). On the other hand dynamic visualizations may overburden the learners’ cognitive capabilities due to large amounts of continuously changing information, particularly if the output variables are represented as non-interactive animations that do not provide learners with the possibility to adjust the playback speed or to watch single frames (e.g., Lowe, 1999). In order to cope with these requirements, learners frequently make use of a strategy that limits their processing to selected aspects of a dynamic visualization, which are often not the most relevant aspects of the visualization, but rather those that are most perceptually compelling (cf. Lowe, 2003). In order to support simulation-based discovery learning it has been suggested to structure the learners’ interactions with the learning environment (e.g., van Joolingen & de Jong, 1991). Typically, these support methods guide learners to focus on specific variables of the underlying model, to generate hypotheses about relationships between these variables, to conduct experiments in order to test the hypotheses, and to evaluate the hypotheses in light of the observed results. Furthermore, various instructional support methods have been developed to facilitate specific processes of discovery learning, such as offering predefined hypothe- ses or providing experimentation hints (e.g., Leutner, 1993; Njoo & de Jong, 1993; Swaak, van Joolingen & de Jong, 1998). However, empirical results regarding these methods of instructional guidance are ambiguous (cf. de Jong & van Joolingen, 1998). Learners frequently did not make sufficient use of the instructional support to increase their learning outcomes. One way to explain these findings is that learners lack prior knowledge necessary to benefit from complex visuali- zations. Learners who do not know enough about the domain of the visualized and simulated concept have problems processing complex dynamic visualizations and to interact with them in a goal-oriented way, even if they have enough information about useful learning behavior (cf. Leutner, 1993; Lowe, 1999; Schauble et al., 1991). Another reason – which is not independent from prior knowledge – is the difficulty of interconnecting multiple representations. Usually, simulations are embedded in multimedia learning Abstract Discovery learning with computer simulations is a demanding task for many learners. Frequently, even fostering systematic and goal-oriented learning behavior does not lead to better learning outcomes. This can be due to missing prerequisites such as the coherent mental integration of different types of representations comprised in the simulations and in the surrounding learning environment. Prior studies indicated that learning performances can be enhanced by encouraging learners to interactively and externally relate different static sources of information to each other before exploring dynamic and interactive visualizations. In an experimental study addressing the domain of mechanics it was largely confirmed that the active external integration of representa- tions can improve simulation-based learning outcomes. Introduction Computer-based learning environments increasingly comprise simulations in terms of dynamic and interactive visualizations to illustrate complex processes and abstract concepts. These simulations may be highly interactive in that they allow learners to change input variables by entering data or by manipulating visual objects and to observe the consequences of these changes in the dynamic visualizations as well as in additional representations such as numeric displays, formulas or text labels. The conceptual model underlying the simulations has frequently to be inferred by the learners in processes of discovery learning, which correspond to the steps of scien- tific reasoning: defining a problem, stating a hypothesis about the problem, designing an experiment to test the hypothesis, carrying out the experiment and collecting data, evaluating the data, and (re-)formulate a hypothesis. The use of simulations frequently aims at inducing active learner behavior and constructive learning processes (e.g., de Jong & van Joolingen, 1998; Rieber, Tzeng & Tribble, in press). Learners have to self-regulate their learning behavior in order to discover the underlying conceptual model, which is assumed to lead to the acquisition of deeper domain knowledge (e.g., Schnotz, Boeckheler, & Grzondziel, 1999). However, it has shown that learners encounter difficulties in all phases of the discovery learning process. For example, learners have problems formulating useful hypotheses, designing appropriate experiments, and evaluating the output variables adequately (e.g., de Jong & van Joolingen, 1998; Njoo & de Jong, 1993; Reigeluth & Schwartz, 1989; Reimann, 1991). Moreover, many learners have difficulties in planning their experiments in a systematic and goal-