Laboratory and Real-Life Research Advance Intelligent Choices
Learning Sciences & Technologies (LST) graduate students follow the WPI method of integrating theory with practice on campus and in urban classrooms, researching the impact of educational technology with input from teachers and students whose learning experiences—and futures—will be enhanced by these efforts. Faculty and students collaborate with the Worcester Public Schools, applying and testing theories in computer science, cognitive and educational psychology, and statistics in city classroom settings.
Working closely with undisputed leaders in educational psychology and educational data mining, as well as faculty at the cutting edge of intelligent tutoring, cognitive psychology, and artificial intelligence, LST graduate students focus on methods to increase success in STEM (science, technology, engineering, and mathematics) learning and teaching in the United States in grades K-12.
Researchers use the following resource centers:
- Artificial Intelligence and Intelligent Tutoring Laboratory
- Math, Abstraction, Play, Learning, and Embodiment (MAPLE) Lab
- Advanced Learning Technologies Research Lab
Along with these laboratories, LST cognitive models are used to study student thought and learning patterns to design educational materials, practices, and technologies. Topics include instructional technologies, learning with visualizations and simulations, learner characteristics, human-computer interaction, and machine learning.
Advanced Learning Technologies Research Lab
We create, explore, and analyze the impact of Learning Technologies for STEM focusing on various aspects of human learning and incorporating data from myriad sources and equipment.
Our research spans many areas, including Personalized Learning, Affect and Motivation, Metacognition, Embodied Cognition and Wearable Devices for Active Learning, Educational Games, Intelligent Pedagogical Agents, Students with Learning Disabilities, Cultural Differences in the Design and Implementation of Learning Technologies, and Learning Technologies for the Developing World.
Educational Data Mining Research Group Predicts Outcomes
Large datasets of students’ fine-grained interactions (e.g., student S answers math problem X with answer Y at time T) with intelligent tutoring systems, educational interventions, and massive open online courses (MOOCs) help us explore and improve how learners learn and how teachers teach.
By harnessing methods from machine learning—such as probabilistic graphical models, Markov chains, and deep neural networks—we can develop more accurate predictors of which and when students will succeed, fail, persist, or need help. These predictors can, in turn, serve as the basis for both human-assisted and automated interventions to improve learning outcomes and the personalization of learning.
Machine Perception of Human Learning Group
This group uses machine learning and computer vision to study how learners learn and how they emote while they learn. Particular interests include the training of deep neural networks to recognize students’ facial expressions during learning tasks, and the development of real-time cyberlearning systems that respond instantaneously to learners’ current cognitive, affective, and linguistic needs.
Quantitative Methods in the Learning Sciences
This research group focuses on rigorous quantitative methods such as hierarchical linear models (which is a typical method to use when students are nested inside teachers and teachers are nested inside schools). Other topics include timely learning science issues like structural question modeling, longitudinal data analysis, propensity score matching, regression discontinuity designs, quasi-experimental designs, and advanced topics like principal stratification. Faculty in this group apply (and adapt) statistical methodologies to solve the problems they are working on.
ARRS Project Assesses Student Understanding
Professor Beck leads the National Science Foundation-funded Automatic Reassessment and Retention System (ARRS) project, which evaluates student topic understanding. Rather than repeat the process of learning something only to quickly forget it, this research aims at periodically reassessing students on the topic.
Earlier studies showed better knowledge retention with the latter approach, and current LST research furthers the creation of a system that will study how to optimize student learning through this method for better retention.