Study: Preschoolers use statistics to understand others

Young children are natural psychologists, says Cornell cognitive psychologist Tamar Kushnir. By the time they're in preschool, they already understand a lot about other people's inner mental lives -- their desires, preferences, beliefs and emotions. But how do they acquire this understanding? In part by using statistics, reports a new study led by Kushnir.

The research, described in the August issue of Psychological Science, a journal of the Association for Psychological Science, "provides the first evidence that young children can use intuitive statistical abilities to infer a psychological cause -- a preference," says Kushnir, assistant professor of human development and director of the Early Childhood Cognition Laboratory in the College of Human Ecology.

For example, in one experiment, preschool children saw a puppet named Squirrel remove five toys of the same type (all blue flowers) from a container full of toys and happily play with them. What varied for different children, however, were the contents of the container. For one-third of the children, all the toys were blue flowers. For another third of the children, only 50 percent were blue flowers; the others were red circles. For the last third of the children, 18 percent were blue flowers and 82 percent were red circles.

When later asked to give Squirrel a toy that she liked, the children were most likely to give her blue flowers when the container held only 18 percent blue flowers, and slightly less frequently when the container had 50 percent blue flowers. When the container had 100 percent blue flowers, they gave her toys at random.

"That means children inferred that the puppet had a preference for blue flowers if the sample of five toys didn't match the proportion of toys in the population (the container)," explains Kushnir. "This is a statistical phenomenon known as non-random sampling."

In another experiment, 18- to 24-month-old children also learned about the preferences of an adult experimenter from non-random sampling. They watched the adult choose five toys -- either from a box with 18 percent of that toy or 82 percent. The adult played happily with the toy either way, but the toddler only concluded that the adult had a preference if she had chosen the toys from a box in which that toy was scarce.
"Our results support the intriguing conclusion that statistical inference plays a critical role in early social learning -- both as infants form initial notions of psychological causality and later as preschoolers achieve more detailed and sophisticated psychological knowledge," write the authors, who include Fei Xu of the University of California-Berkeley and Henry M. Wellman of the University of Michigan.
Early statistical intuitions are not conscious or explicit, Kushnir said. "Xu previously found that infants have expectations about random sampling when they are 8 months old. What we did was ask -- could these expectations be useful for learning about other people? We found that indeed they could. When infants' expectations are violated they look for a hidden cause. In this case, the hidden cause is a preference for one toy over another.

"Babies are incredible natural learners," adds Kushnir. "Babies and children are like little scientists. They gather evidence by observing and experiencing the world. Later on, there will be time for formal instruction, but when they're really young, this sort of informal learning is critical."
The research was supported by the McDonnell Collaborative Initiative on Causal Learning, the National Institute of Child Health and Human Development and the Natural Sciences and Engineering Research Council, Canada.

This story is adapted from an article by the Association for Psychological Science.

Media Contact

John Carberry