Coupling Models With Experiments to Support Knowledge Generation

• In Event: 63.025 - Supporting Modeling Epistemologies in the Science Classroom

Abstract

Objective. Scientific models are epistemically generative, yet science curricula often position models as simply representing knowledge. It is rare that students use models to generate new ideas or questions (Schwarz et al., 2009). We designed “hybrid” labs that couple exploration of computational models with laboratory experiments. Students use both methods to explore a larger research question. They are encouraged to use the computational model both to inspire possible experiments and to extend and contextualize their experimental results. In this research we investigate how this design can support generative uses of models together with experiments.

Theoretical framing. Models are often positioned below experiments on the epistemic hierarchy (Morrison, 2017). For example, models are often described producing experimentally testable hypotheses, while experiments ultimately generate knowledge by confirming or rejecting such hypotheses. In scientific practice, modeling does more than identify hypotheses; models are used to explore variation and dyanmics in system behavior, draw attention to the boundaries and transition states, and generate output for independent analysis (Odenbagh, 2005). In these roles, models do not simply serve experiments; instead experiments and computational modeling can be used together as part of a larger methodological system to support knowledge generation (MacLeod & Nersessian, 2013).

Approach. In this study we explored how undergraduate students in a “hybrid” lab course framed the relationship between models and experiments and how this epistemological framing (Elby & Hammer, 2010) influenced the extent to which they engaged in knowledge generation. We analyzed in-class video to characterize how students used the computation model together with experiments. We also analyzed interviews to triangulate our understanding of how students were framing the relationship between the modeling and experimental methods. We report on a comparative case study of two students whose use of the computational model reflected different framings of the relationship between the model and experiments. In one case, a student was able to generate theories and evidence by exploring the model as a generative tool that complemented and extended his experimental results. In the second case, a student did not use the model to generate new ideas because she framed the simulation as functioning only to recreate her experimental results.

Implications. We highlight two implications that have emerged from this work. First, we propose that one potentially productive way to foreground the generative power of models is to position them as coupled with, rather than subservient to, experiments. The design of the lab curriculum positions the simulation and experiment as alternative methods to explore questions. It allows students to move between the two, filling in the gaps in one method with outputs from another and to integrate knowledge from each. Second, we suggest that more work will be needed to help students take up this framing. In particular, instructors must encourage the idea that computational models have relevance beyond representation of system features or the replication of experimental results.

Authors

• Katherine Anne Parker, Tufts University

• Aditi Wagh, Tufts University

• Julia Svoboda Gouvea, Tufts University