The Role of Mechanistic Explanations in Learning about Science and Technology

Yale University, New Haven CT

Investigators

Abstract

This project, led by a team of researchers from Yale University, will focus on the role of causal mechanistic explanations in children's and adults' learning of STEM content and will examine the relations to engagement with science education. Learning the specific mechanistic details of natural phenomena and devices is often envisioned as an ideal goal of STEM learning, yet such information is typically forgotten soon after instruction by children, as well as by adults. The researchers will pursue the hypothesis that such mechanisms should still be taught because, after the specific details have been forgotten, what remains promotes enduring representations and facilitates higher order learning. This alternative view emphasizes learning that much more closely mirrors how informal science (and often formal science as well) actually works. This project will build on recent cognitive science discoveries about the kinds of knowledge that are most robustly represented both in folk science and in formal science. Four sets of experiments will explore developmental changes in preferences for mechanistic explanations, what is retained from exposure, the effects of different learning goals on the use of mechanistic information in learning STEM content, and the development of abilities to infer internal mechanistic structures. If the hypothesis is supported, the results will suggest that exposure to causal mechanistic information is essential to early science education and that such content should not be omitted even though its details are forgotten. The project is funded by the EHR Core Research program, which supports fundamental research that advances the research literature on STEM learning, and has implications for education in both formal and informal settings. This project will involve four sets of randomized controlled experiments, primarily with children at grades K, 2, and 4. Study Set 1 will explore the cognitive basis and development of children's preference for explanations that involve a causal mechanism. The researchers will use a choice paradigm that contrasts mechanistic explanations with other forms of information. Children will be asked which explanation they would like to learn more about for a given object. Strong preferences for mechanistic information are predicted at all ages, but with developmental differences relating to overall complexity, the domains queried, and kinds of mechanism involved. Study Set 2 will present children with mechanistic explanations for devices and biological entities. Children will then be assessed for retention of mechanistic details, functional properties, non-mechanistic details, judged complexity, broad causal patterns such as causal centrality and potency, and which of two experts to consult to learn more about the entity. These children will be compared to control children who will also learn about the same internal parts but with no causal or functional language that reveals mechanism. Rapid decay of detailed mechanistic information is expected in both groups, but the mechanism group should show more enduring memories for information relating to complexity, centrality, and expertise domains. Study Set 3 will focus on goal framing. The researchers will first look at children's and adults' intuitions about what is retained after exposure to causal mechanistic explanations, expecting a strong explicit bias favoring retention of vivid details and a neglect of what actually endures and is most often used. Later studies will ask how different learning goals interact with mechanism instruction. Greater enjoyment, engagement, feelings of accomplishment and retention of information are predicted with learning goals stressing more cognitively feasible and useful outcomes. Study Set 4 will focus on complexity intuitions. The investigators will look at how internal complexity is inferred for devices and animals based on behavioral diversity. They will also explore how mechanistic complexity is instantiated in different domains.