GCR: Towards a Convergent Understanding of the Dynamics of Uncertainty In Individuals and Groups with a Focus on STEM Education

Tufts University, Medford MA

Investigators

Abstract

Ambiguity, uncertainty, and confusion (AU&C) are inherent in human experience, at small scales and large, in everyday social interactions and in professional expertise, in activities ranging from students collaborating on a physics problem to a team of EMTs responding to an accident or a physician determining a diagnosis from imprecisely communicated symptoms. Despite its prevalence, we lack systematic understanding as to how people engage with AU&C. This is not surprising. The ways in which humans manage these processes as individuals and in groups involve highly complex phenomena lasting from minutes to years, characterized by changing information, social context, and immediacy. Studying AU&C by collecting and analyzing data in controlled and natural settings is currently not possible at the scale required to make progress in this domain. Addressing this gap is a Grand Challenge requiring the insight of experts from the social sciences, data sciences, and engineering. This project focusses on AU&C in the context of science, technology, engineering, and mathematics (STEM) education. Current educational practices frame AU&C as liabilities to avoid, thereby inducing stress and often behavioral impasses when encountered. Laboratory and classroom research have shown that AU&C can be framed as an exciting and motivating challenge with additional positive effects on learning traditional content. Motivated by such results, these researchers will study the dynamics of AU&C during STEM learning in individuals, small groups, and classes over broad time scales. This project will provide fundamentally new capability and insight into the dynamics of AU&C in STEM environments Bringing together experts in Learning Science, Cognitive Science, Modeling Sciences (math, signal processing, statistics, and machine learning), and Engineering (systems and sensors), these researchers will undertake a convergent approach to understand productive engagement with ambiguity, uncertainty, and confusion (AU&C) as a target in STEM learning and encourage engaging with AU&C in problem solving. The team will develop methods for collecting and analyzing longitudinal data from a diverse population of students using a multimodal data approach that includes behavioral, linguistic, and physiological sensing. Advanced wearable sensors will be developed to inconspicuously collect data for long periods of time, thus allowing for observation of the impact of AU&C on physiological states. Lab studies will examine AU&C in problem solving in controlled settings by individuals and groups. Data from the classroom experience of a larger group of students that includes the lab cohort will be collected for their first two years. The researchers will build computational systems adapted to the needs of storing and accessing large heterogenous collections of human subjects’ data. Working collectively across domains, they will construct state of the art machine learning and natural language processing methods to interpret these data and model AU&C dynamics at the individual, small group, and classroom scales within STEM learning environments. Ultimately, this project will create new methods to study real-time, in-context learning using large language and dynamical models to fuse multimodal data. Results from this work promise to revolutionize how students and trainees learn to recognize and engage productively with complex AU&C. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.