RU-N Professors Awarded $750K NSF Grant to Study How Kids Learn

NSF Science of Learning Collaboratives grant involves six researchers spanning three disciplines at five universities.
 

Two RU-N professors have been awarded a $750K National Science Foundation (NSF) grant to study the conditions under which children learn best, drawing on research from the fields of computer science, psychology and education.

Patrick Shafto, an Associate Professor of Mathematics and Computer Science who holds the Henry Rutgers Term Chair in Data Science, and Liz Bonawitz, an Assistant Professor of Psychology specializing in cognitive development and childhood learning, are co-Principal Investigators (PIs), leading a cohort of six researchers spanning five universities for the two-year study.

The grant is part of NSF’s Science of Learning Collaboratives program, aimed at building bridges across disciplines to answer big questions that advance our knowledge about learning.

“The work Liz and I do is already interdisciplinary, and we’ll be working with some of the foremost researchers in these fields and bring all of our perspectives to bear on some big questions,” says Shafto.

Shafto runs a Cognition and Data Science Lab at RU-N, which is developing tools to facilitate human learning and decision-making, while also advancing machine learning. Bonawitz blends cognitive psychology with computational modeling to get at how kids learn, and what drives their curiosity and engagement.

They’ll work with colleagues at Boston University, UC Berkeley, Temple, and the University of Delaware who specialize in psychology and education, some of whom also add computational modeling to their arsenal.

At the heart of the team’s research is a longstanding debate on whether children learn best from direct instruction, free exploration or a middle-ground called guided learning, in which adults hold a curricular goal but allow children to lead and explore, while building on the kids’ focus.

Prior research suggests the latter method is the most effective, but Shafto insists this needs to be pushed further.

“We’re hoping to get at the when, why and how guidance can foster children’s desire to learn, providing empirical evidence and then seeing how this plays out in actual classrooms,” he says.

They’ll look at when children are inclined to explore, the role of guidance for the success of exploratory play, the inferential implications of instruction on learning; developmental and other considerations related to learning from others, and computational models that predict the strengths and limitations of learning from instruction.

Shafto sees the value in approaching these questions from multiple perspectives, adding that he can’t possibly appreciate all the concerns that in cognitive scientists and education researchers bring to the table.

Bonawitz concurs.

“My cognitive science work is missing application. We need to know how the basic science I’m doing impacts actual education, and how all this comes together requires a strong computational framework,” she says. “So, having all three facets—computational and empirical science connected to education—is important.”

The team will start by giving talks at conferences, then bringing their individual research together to design experiments and build from there. Though they’re at the beginning of their multi-year project, they’re sanguine about its prospects.

“The whole point of this collaboration is, how can we do this in a way that’s satisfying, so we can make scientific claims that are relevant to actual learning environments?” says Shafto. “These are hard problems worth working on. They’re what motivates the entire team.”

Top inline photo: By Lawrence Lerner
Bottom inline photo: Courtesy of Liz Bonawitz
Header photo: public domain image