Augmenting Spatial Thinking With Molecular Models in STEM

• In Event: 25.017 - Supporting Spatial Thinking to Enhance STEM Learning

Abstract

Students in Organic Chemistry must master many different diagrammatic representations of molecules, which is very challenging for students. Recent research shows that manipulating 3-D molecular models can facilitate understanding of these representations (Padalkar & Hegarty, 2015; Stull & Hegarty, 2016; Stull, et al, 2012) by augmenting students’ spatial thinking and addressing their misconceptions.

Spatial representations are essential to communication, research, and teaching in STEM disciplines, such as Organic Chemistry where they are the lingua franca, yet they place a high cognitive demand on students, especially those with less spatial ability. We propose three ways in which manipulating 3-D models support students’ spatial reasoning. First, manipulating models enables students to off-load cognition on to external objects and performed actions (Kirsh & Maglio, 1994) thereby reducing demands on working memory (Wu & Shah, 2004). Second, the use of multiple representations helps students build more complete mental models (Ainsworth, 2006) because different representations make salient different aspects of the referent. Third, 3-D models help students integrate new information with their preexisting knowledge (elaborative encoding) by supporting enactment (Cohen, 1989) and offering multiple perceptual modalities (Barsalou, et al., 2003) to support improved memory encoding and recall.

Seven studies used a quantitative approach to compare model-supported learning against a no-model control under different support conditions. Participants were undergraduate students at a research university. The task in each study was to translate a molecule represented in one form of structural diagram into a different type of diagram of the same molecule. The first set of studies investigated if and how students spontaneously used models to support their reasoning when translating between different structural diagrams (Study 1) and whether encouraged use of models increased model use (Studies 2 and 3). The second set of studies developed and tested a model-based feedback intervention to encourage model use and to improve translation accuracy (Studies 4 and 5). The third set of studies compared usability with physical and virtual models and assessed whether model-based feedback contributed to long-term learning (Studies 6 and 7).

The results in all seven studies are consistent in demonstrating that students are more successful in translating between diagrams when models are available, that enacting translations with models is a better predictor of learning than a student’s spatial ability, and that a model-based feedback intervention dramatically improves learning. In addition, model-based feedback is superior to verbal feedback alone, models scaffold learning rather than act as a crutch, learning with model-based instruction is resilient over a delay of several days, and learning with models transfers to performance when models are no longer available. Finally, concrete models are equivalent to virtual models in promoting learning.
These studies are significant in suggesting that models (a) augment spatial thinking by externalizing spatial representations that would otherwise drain cognitive resources for learning, (b) confront students with their illusions of understanding when working with spatial content, and (c) support enactive learning by allowing students to represent processes that would otherwise be difficult to imagine.

Authors

• Andrew T. Stull, University of California - Santa Barbara

• Shamin Padalkar, Tata Institute of Social Sciences, Mumbai, India

• Mary Hegarty, University of California - Santa Barbara