Research may provide the tools to create better schools

Just as clinical trials are critical to enhancing human health and medicine, field experiments are critical to understanding human learning and education, according to a new paper published in the online journal Science.

A team of economists and psychologists at Harvard and Massachusetts Institute of Technology (MIT), working together with organizations focused on enhancing education and reducing poverty in India, have demonstrated both the feasibility and the necessity of such experiments, the study noted.

“Research in cognitive science has taught us a great deal about what children know and how they learn,” said Harvard Professor Elizabeth Spelke, the Marshall L. Berkman Professor of Psychology and one of the paper’s authors. “Much of that work applies to all children, worldwide, and it gives us methods for assessing children’s knowledge that are robust enough to work when implemented by adults with only high school education, working with children in hot and noisy slums. This research doesn’t tell us how to create better schools, but it gives us the tools to do field experiments that can.”

While primary education now is compulsory in India, it was not widespread a generation ago. Poor Indian children are likely to live in families and communities with high rates of illiteracy and innumeracy. Countrywide studies show that many fail to learn the basic concepts and skills taught in primary schools, in large part because the curricula taught in the schools assumes a level of preparation that is not there.

Drawing on decades of cognitive science research probing the nature and early development of mathematical reasoning, the Harvard and MIT team created a program for enhancing poor children’s readiness to learn mathematics, and evaluated it over a period of 18 months. More than 200 single-class preschools serving 1,500 children in Delhi were randomized to three conditions. One group received a math curriculum consisting of games training two aspects of intuitive mathematics: sensitivity to numbers, and geometry. Children from wealthy countries master these intuitive mathematics over their preschool years, and research in cognitive science suggests this is critical for later learning of symbolic mathematics.

A second group received a curriculum of games with the same structure that trained sensitivity to aspects of human communication: sensitivity to emotional expressions and signs of attention. This skillset also develops in preschool and is widely thought to be critical to learning from others, but was not expected to affect math directly. Over a four-month period, the games were played for three weekly one-hour sessions in these two groups, while a third group of children received the regular preschool curriculum.

To measure the effects of the games, all the children were assessed on a variety of abilities, including intuitive abilities close to the content of the games in the two curricula, such as determining which of two sets of objects was more numerous or which of two faces was happier, and symbolic abilities at the center of the primary school curriculum, such as identifying Arabic numerals or naming shapes.

In the summer months immediately following preschool, the children showed effects of the math games intervention that closely paralleled effects found in Western children from developed countries: higher sensitivity to numbers and geometry on the intuitive measures that were close to the games, and also better mastery of the language and symbols of intuitive, preschool mathematics. In these respects, the study strongly confirmed the central findings from basic research in the cognitive science of mathematics, showing that those findings generalize across children living and learning in very different circumstances. Moreover, they show that field interventions can be effective, not only when implemented in carefully controlled model classrooms but also when implemented and evaluated by adults with minimal training and little connection to the research team.

Shortly after the end of the first year of primary school, the children who had received the math games curriculum still reliably outperformed the other children at tests of intuitive numerical and spatial abilities. Although these children had no access to the math games after the end of the intervention, the benefits from the games persisted.

In striking contrast, the intervention showed no effect on children’s mastery of symbolic school mathematics: By the end of first grade, children in the math games condition were no better than those in the other two conditions at deciphering Arabic numerals, performing simple verbal additions, or learning the vocabulary of school geometry. Just as well-supported findings in molecular biology do not guarantee the success of new medical treatments, key findings in cognitive science were reproducible in this population but failed to produce a successful educational curriculum.

“We think our findings underscore both the feasibility and the necessity of randomized-controlled field experiments to test frontier cognitive science hypotheses in the field,” said Spelke.

Why did the curriculum fail to improve school mathematics, even as it effectively trained intuitive concepts, and how might future studies succeed at this second step?

“Our best guess is that our intervention did not increase kids’ performance in school because the gap between intuitive mathematics and formal mathematics is too large, and ordinary conversations and social interactions cannot bridge it in this context,” said Esther Duflo, a professor of economics at MIT and a co-author of the study.

“Our ongoing interventions focus on versions of these games that exercise children’s intuitive mathematical abilities while also presenting the primary language and symbols of school mathematics. We should soon learn whether the new curriculum works better. In this respect, field research is no different from basic research: Both require multiple experiments, and learning from failures, to get things right.”

This research was funded by the UBS Optimus Foundation and by Harvard’s Foundations of Human Behavior Initiative.