Providing Guidance and Opportunities for Self-Assessment and Transfer in a Simulation Environment for Discovery Learning

Providing Guidance and Opportunities for Self-Assessment and Transfer in a Simulation Environment for Discovery Learning Jason Tan 1 , Nathan Skirvin 1 , Gautam Biswas 1 , Kefyn Catley 2 {jason.tan, nathan.skirvin, gautam.biswas, kefyn.catley }@vanderbilt.edu Department of EECS / ISIS, Vanderbilt University Department of Teaching and Learning, Vanderbilt University Nashville, TN 37235 USA to answer questions and solve problems in different situa- tions. Researchers have confirmed that “exploratory learn- ing” with understanding and the ability to transfer in learn- ing environments can lead to “effective learning” (Schwartz and Bransford, 2005). In our work on middle school science education, we have been developing a teachable agent environment called Betty’s Brain, where students teach the computer agent, Betty, by creating concept maps. The concept map represen- tation and reasoning mechanisms are geared towards teach- ing and learning about interdependence among entities in river ecosystems. Analysis of student answers to post-test questions on balance (equilibrium) made it clear that stu- dents did not quite grasp this concept and how it applied to river ecosystems. We realized that to understand balance, students had to be introduced to the dynamic behavior of river ecosystems. Analyzing dynamic systems behavior can be very chal- lenging for middle school students who do not have the relevant mathematical background or the maturity to grasp these concepts. To scaffold the process of learning about temporal effects, we have developed a simulation-based learning environment that supports scientific discovery learning in the domain of river ecosystems. This support comes in the form of guidance, scaffolding, and mecha- nisms for self-assessment and transfer. The addition of a simulation environment allows students to explore and con- duct experiments about dynamic processes in the ecosystem. Abstract This paper describes the use of a computer-based simulation learning environment to teach fifth grade students about dy- namic processes in a river ecosystem using guided discovery learning techniques. The simulation is framed in an estab- lished learning-by-teaching environment called Betty’s Brain, where students create causal concept maps to understand in- terdependence and balance. Students were asked to manipu- late and observe the simulation and teach the computer agent, Betty, using the knowledge acquired from the simulation. We present the design of the simulation environment and the im- plementation of its components: the simulation engine, the display and control interface, tools for guided discovery learning, and the self-assessment subsystem. A study was run on two classrooms to examine the effectiveness of the system. Student understanding was measured by pre to post test dif- ferences. Students using the system showed significant learn- ing gains in important concepts. We also studied student learning behaviors in the simulation environment, and found that those who better utilized the self-assessment system per- formed better on the post test. Keywords: simulations, scientific discovery learning, self- assessment, transfer, learning by teaching Introduction Constructivist approaches to education involve experiences through observation, exploration, and performance (Dewey, 1938). Tools provided as scaffolds can support learners in constructing their own knowledge with less frustration, more motivation, and more efficiency and innovation in the learning process. These tools can enhance activity and thinking (Vygotsky, 1978). Computers have become a ver- satile tool through which one can gain a wide variety of learning experiences. In particular, the integration of simu- lation, graphics, and animation enables users to experience and witness processes that might not otherwise be readily observable and comprehendible. This makes computer- based simulations a powerful tool for learning. In simulation-based learning environments “the main task for the learner is to infer the characteristics of the model un- derlying the simulation” (de Jong and van Joolingen, 1998). In other words, the simulation environment provides learn- ers with observations and experiences that they must at- tempt to explain, assimilate, and combine with their existing knowledge. One way to do this is to provide scaffolds and guidance that help users generate hypotheses, use interactive controls to manipulate the simulation, run experiments and verify the hypotheses, and then apply the learnt knowledge Scientific Discovery Learning Scientific discovery learning environments provide the learner with resources and tools that help them deduce prop- erties of a specific domain by running controlled experi- ments or by accessing relevant data that is made available in the environment (van Joolingen and de Jong, 1997). They promote a constructivist form of learning where students start with a science problem, formulate hypotheses, design experiments, and then discover the relations needed to solve the problem. The learning environment may include addi- tional resources and tools to help students focus on the pri- mary entities and relations that are important to the discov- ery learning process (de Jong and van Joolingen, 1998). Simulations can form the core of scientific discovery learning. They represent models of the underlying phenom- ena or domain of study (Wilensky, 2006). The primary task for the student is to run experiments with the simulation and inspect the observed behaviors to discover the model or