Seeds of Hybrid Modeling in Design-Oriented STEM

• In Event: Thinking Across Representations: Designing to Support Investigations That Leverage Physical and Computational Modes of Inquiry

Abstract

Objectives. In the context of K-12 STEM, we aim to identify “seeds” (Lehrer & Schauble, 2015, p. 678) of hybrid modeling. Hybrid modeling involves creating multiple models that parallel different aspects of a phenomenon (Nersessian, 2017). Our analysis explores how 6th graders reason with and across physical and virtual models that represent population dynamics and the survival needs of plants and animals.

Theoretical framework. Professional scientists make progress on ideas by reasoning across representations, including physical and computational models (e.g., Gooding, 2006; MacLeod & Nersessian, 2013; Nersessian, 2017). When research aims are descriptive, tightly coupled representations allow scientists to identify discrepant events and triangulate data (MacLeod & Nersessian, 2013). Research in science education has explored how tightly coupled experimental-computational systems could support students’ learning in the context of descriptive tasks, like identifying phases of bacterial growth or considering when it is advantageous for bacteria to have high or low mutation rates (Author, 2016; Author, 2018). These tasks often afford direct mapping between systems, because the scale is the same for the experimental and computational systems.
STEM can also involve design-oriented components. Nersessian (2017) describes the dual objectives of biomedical engineers, who aim to contribute to “basic science” while designing novel technologies. For biomedical engineers, direct mapping between models and target phenomena is often impossible due to issues of ethics, scale, or control. To approximate salient features of phenomena, biomedical engineers create multiple models that parallel different aspects of phenomena. Similarly, in K-12 science, some topics cannot be investigated directly due to issues of ethics and scale.

Research Question: How do students leverage physical and computational models to make progress on a task that blends learning and design while managing constraints related to ethics and scale?

Methods and data. This work is part of a design study (Cobb et al., 2003) aiming to integrate computational modeling in an NGSS-aligned unit (NGSS Lead States, 2013). The 9-week unit is anchored in explaining differences between guppy populations. Students design a closed-system biosphere to test ideas about what guppies need to survive. They also build computational models to consider levels of predation and competition.
To identify seeds of hybrid modeling, we follow two students, Caleb and Karla, who frequently reason across physical and virtual representations to explore ideas. We analyze interviews, whole class videos, and videos that capture students’ interactions with their computers. We compare students’ practices to the hybrid modeling practices of biomedical engineers (Chandrasekharan & Nersessian, 2017; Nersessian, 2017).

Results and significance. We find that Caleb and Karla frequently engage in hybrid modeling practices, including: (1) creating increasingly complex hybrid devices, (2) balancing dual conceptual and design objectives, and (3) leveraging models for discovery and transfer (Table 1). These seeds of hybrid modeling can be used to inform designs for learning, particularly in design-oriented contexts with constraints related to ethics and scale. This is an important contribution because philosophers of science (e.g., Fox Keller, 2003; Nersessian, 2017) have identified hybrid modeling as a feature of emerging scientific fields.

Authors

• Ashlyn Pierson, The Ohio State University

• Corey Brady, Vanderbilt University

• Doug Clark, University of Calgary