Why failure can be productive: Underlying learning mechanisms of problem solving prior to instruction approaches

• Applicant: Professorin Dr. Katharina Loibl
• Subject Area: General and Domain-Specific Teaching and Learning
• Term: since 2020
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 439706396

Learning approaches with an initial problem-solving phase followed by an instruction phase (short: problem solving prior to instruction, PS-I) have received increasing empirical attention in the last decade. The results from PS-I studies suggest that these instructional designs can improve conceptual understanding and transfer. While there are first attempts of working towards a theory on the underlying learning mechanisms of PS-I, the theoretical foundation is still fragmented. The lack of a coherent theory is partly caused by the fact that the different mechanisms under discussion are not well confined from each other, are not connected to related theories, and are not tested systematically. Therefore, the network pursues three strategies in order to work towards a coherent theory:1. Systematizing the PS-I findings:We systematize the existing studies in order to derive a concise picture of the findings on PS-I and to identify boundary conditions for beneficial learning outcomes. For example, multiple studies have shown beneficial effects of PS-I in mathematics and physics, whereas there is only a limited number of studies in other domains and these studies show mixed results. Against this background, a relevant question is whether PS-I is only suitable for specific types of learning goals. Similar questions are triggered by other boundary conditions (e.g., age: What preconditions of the learners are necessary?).2. Connecting the PS-I literature to related theories and findings:While sorting the PS-I literature may help to identify boundary conditions, the existing PS-I literature cannot explain these boundary conditions: Why is PS-I suitable for specific types of learning goals? Why may certain preconditions influence the effect on learning? Connecting the PS-I literature to related theories and findings (e.g., discovery learning, worked examples, collaborative learning, KLI framework) may help answering such questions. For instance, ‘encoding of principles’ is discussed as one of the core learning mechanisms that render learning from worked examples effective. Similarly, PS-I supports students in recognizing deep features of the target concept. It seems reasonable to argue that the terms ‘deep features’ and ‘principles’ have similar meanings. Therefore, we take the theory of example-based learning into account when working towards a theory on the underlying learning mechanisms of PS-I. Similar parallels can be found in relation to other theories and research strands.3. Testing the theoretical assumptions:Finally, the derived theoretical assumptions should be tested systematically. Currently, different research groups work on PS-I, and the conducted studies of these research groups vary in multiple ways. In consequence, it is impossible to derive clear implications from a synopsis of these studies. A more synchronized research program would allow testing the theoretical assumptions of the mechanisms underlying PS-I more rigorously.

• DFG Programme: Scientific Networks