Modeling as a way to promote and reveal learning about biological systems

Background/Question/Methods

Learning sciences research suggests that engaging students in the core epistemic practices of disciplinary science can promote deeper conceptual understanding and greater fluency with disciplinary constructs compared to traditional (aka, lecture-based) instruction. However, large enrollments typical of many introductory biology classrooms can pose challenges for implementing instruction and assessments that reflect the skills and practices of contemporary biological research.  Models are a pervasive construct throughout STEM disciplines that facilitate communication and analysis by distilling volumes of information into manageable units.  Biologists regularly use models for constructing and testing hypotheses, evaluating evidence, warranting arguments, and identifying system unknowns.  As such, models are foundational in the practice and epistemology of biological science.

We developed an instructional approach that uses system models as a way to teach, learn, and assess students’ understanding about biological systems.  Our adaptation of a system model derives from Structure-Behavior-Function Theory in which structures (model components, represented in boxes) are linked by behaviors (arrows representing relationships between pairs of structures) that collectively describe a system’s function (role or purpose).  Collectively, our research explores how system models can be used to reveal student thinking and promote the development of skills necessary for learning and reasoning about complex biological systems. 

Results/Conclusions

Our work with models in college-level biology suggests:  (1) Models are useful for documenting change in student thinking.  We analyzed student-constructed models at three times during a 15-week semester of majors introductory biology (n=368). While correctness increased throughout the semester (p<.001), complexity (degree of connectedness) peaked at midterm (p<.001), then declined toward the final exam (p<.02).  We hypothesize students were converging toward more parsimonious models containing fewer instances of irrelevant information and more biologically accurate language.  (2) Lower-achieving students may derive the greatest benefit from a model-based pedagogy.  We compared performance across achievement tritiles and observed that lower-achieving students showed the greatest relative gains, reducing their performance gap with high-achievers from 21% to 13%.  (3) Models facilitate metacognition by helping students organize and reflect on their thinking. We compared students’ model vs. essay responses to a common assessment.  Although students overwhelmingly preferred constructing their model prior to writing their essay, indicating it “was easier” or “helped them organize their thoughts”, we found little evidence of overlap in concepts or ideas across assessment formats.  In our symposium, we will discuss research from the cognitive and learning sciences that underpin our findings, as well as implications for instruction and best practices.