Better Data on How Babies Learn

“I have spent the past decade trying to understand the mystery of how children learn so much from so little, so quickly,” says Laura Schulz, the Charles F. Hopewell Professor of Cognitive Science, introducing her 2015 TED Talk. She is also a principal investigator of the Early Childhood Cognition Lab (ECCL).

As Schulz demonstrates in one study, babies are adept at drawing accurate inferences—making rich generalizations from small samples of data. But in order to make inferences in scientific studies, developmental psychologists could sometimes benefit from the opposite. “Babies learn from sparse data but we could learn a lot more about babies if we had access to big data,” as she puts it. In September, the ECCL rolled out an updated version of “Lookit” (a beta version was tested in 2014), an online research laboratory conceived and implemented by ECCL graduate student Kim Scott, that makes this kind of data collection possible.

In the pilot study currently hosted on the site, a baby watches a pair of videos in which something normal—a ball rolls off the table—is shown side by side with something strange—the ball falls up. Reactions, such as where babies focus their attention, are recorded by researchers via webcam and can give insight into what a baby intuits about physics. Parents from all over the world can participate from the convenience of their own homes, dramatically widening the ECCL’s demographics.

“Lookit addresses the participant bottleneck in developmental research,” says Schulz. “It’s hard to get data sufficient to look for individual differences or test the predictions of quantitative models. Lookit has the potential to expand both the people we can reach and the questions we can ask.”

Baby and parent participants in the "Lookit" online lab. Courtesy of the researchers.