Learning From Worked Examples, Erroneous Examples, and Problem Solving: Toward Adaptive Selection of Learning Activities

摘要

Problem solving, worked examples, and erroneous examples have proven to be effective learning activities in Intelligent Tutoring Systems (ITSs). However, it is generally unknown how to select learning activities adaptively in ITSs to maximize learning. In the previous work of A. Shareghi Najar and A. Mitrovic, alternating worked examples with problem solving (AEP) was found to be superior to learning only from worked examples or only from problem solving. In our first study, we investigated whether the addition of erroneous examples further improves learning in comparison to AEP. The results indicated that erroneous examples prepared students better for problem solving in comparison to worked examples. Explaining and correcting erroneous examples also led to improved debugging and problem-solving skills. In the second study, we introduced a novel strategy that adaptively decided what learning activity (a worked example, a 1-error erroneous example, a 2-error erroneous example, or a problem to be solved) is appropriate for a student based on his/her performance. We found the adaptive strategy resulted in comparative learning improvement in comparison to the fixed sequence of worked/erroneous examples and problem solving, but with a significantly lower number of learning activities.