Current Research Projects


According to the latest standardized test data, the majority of American students possess shallow understandings of scientific principles. An instructional method that has shown great potential in addressing this problem is called Invention-with-Contrasting-Cases. Students receive contrasting cases that highlight relevant features, plus a directive to Invent a representation of a fundamental scientific structure that accounts for the variation across cases. Evidence from classroom-based research has found that compared to traditional instructional methods, Invention leads to greater noticing of deep principles, enhanced learning and transfer, and greater preparation for future learning. The method has had demonstrative success with middle, high school, and college level classrooms and in several STEM domains. In many ways, the Invention method is ready to scale up. However, one weakness of the method is that students need frequent teacher guidance to make progress at Invention tasks.

To begin the scale-up process, this project will develop a computer-based Invention Tutor to guide students through the challenging and messy process of Invention. The Invention Tutor will simulate the guidance of a well-trained inquiry teacher, who asks critical questions and gives feedback just at the right time, to push students’ thinking forward. By tracking student trajectories through the Invention process, the Invention Tutor will respond adaptively to student moves. Most importantly, the tutor will eliminate the constraint of needing a large teacher: student ratio to implement the Invention method successfully. This work will take us one step closer to scaling up a successful instructional technique, with the ultimate goal of enhancing deep learning and transfer in science domains, for both high and low-achieving student populations. The research will identify effective forms of support for Invention tasks and expand our understanding of the Invention process itself. More generally, findings may inform the field’s understanding of the process of discovery and how best to guide it. In addition, the Invention Tutor will represent a new kind of tutor that guides open-ended discovery tasks, as opposed to the more typical tutor that guides students to solve well-defined problems. Below are the four project aims.

Aim 1: Understand the process of Invention.

Aim 2: Identify productive types of support for Invention.

Aim 3: Develop an Invention Tutor to support students through the Invention process.

Aim 4: Assess and refine the effectiveness of the Invention Tutor.

This is a collaborative effort with Vincent Aleven at Carnegie Mellon University.

RumbleBlocksCreating educational games that are fun and engaging but also promote learning is a real challenge. This work explores how to adapt research-based instructional practices into educational games without killing the element of play. The project was initially launched as part of DARPA’s ENGAGE program, which brought game designers and learning specialists together to design and test effective educational games. One of the games we created is RumbleBlocks, a game that teaches K-3 students about the physics of stable structures. For this project, we adapted a successful instructional method — learning with contrasting cases — for the RumbleBlocks game. Contrasting cases are designed to help learners notice critical elements or deep features of examples. Research explores how students reason with the contrasting cases and whether the contrasting cases elements of the game are effective at enhancing learning and transfer beyond the game.

This work is a collaborative effort with Vincent Aleven at CMU’s Human-Computer Interaction Institute and Scott Stevens and Mike Christel at CMU’s Entertainment Technology Center.