Social psychology has long suggested that children take a “monkey see, monkey do” approach to learning. But new research being done at Yale shows infants’ tendency to imitate may be so hard-wired that watching an adult do something incorrectly can make it harder for a child to do it correctly.

A study published yesterday by Yale psychology researchers found that three- to five-years-olds engage in “over-imitation” when confronted with a novel object. The children not only faithfully repeat unnecessary actions an adult performs on the object, they also use their observations to revise their mental concept of the object, said Derek Lyons, a doctoral candidate in psychology and lead author of the study.

“Even when you add time pressures or warn the children not to do the unnecessary actions, they seem unable to avoid reproducing the adult’s irrelevant actions,” he said. “They have already incorporated the actions into their idea of how the object works.”

The study found that such over-imitation exists even when the children are trained to discriminate needless actions from necessary ones using familiar objects, said Jennifer Lin ’09, a student volunteer who has helped run trials for the study.

Before the experimental trials, participating toddlers underwent a “training period” during which they were shown six to eight objects containing turtles inside the objects. They were then asked to determine whether actions an adult performed on the objects were necessary or unnecessary to extracting the turtles, Lin said.

These trials confirmed that children can distinguish extraneous actions from those necessary for a particular objective, she said. But when the objects became unfamiliar and more complex, Lin said, the children could no longer make this distinction.

In the experimental trial — which involved a transparent three-dimensional box that contained a turtle and was attached to a knob — the children mimicked needless actions they saw adults perform, such as tapping the box and turning the knob, she said.

“The effect is really very counter-intuitive,” Lin said. “The issue is not that they don’t physically understand extraneous actions — baseline trials show that in the absence of the adult model, the kids don’t mimic the unnecessary actions.”

The study challenges the dominant notion in social psychology that children imitate adults in order to fit in or because they think it is appropriate to do so, Lyons said. Children who participated in the study could not provide any explanation for why they imitated the actions they did, which suggests that over-imitation operates automatically, even subconsciously, he said.

Over-imitation also somehow shapes the way children think about how objects work, Lyons said. The group formed this hypothesis based on a second experimental condition in which the toddlers were told the “game” had finished but that the experimenter needed help retrieving the turtles for the next participant. Even outside the experimental situation under time pressure, children continued to perform the extraneous actions, suggesting that they had encoded them as essential to working the object, Lin said.

Lyons’ group of researchers based its study on a finding by psychologists Victoria Horner and Andrew Whiten suggesting that chimpanzees and infants learn differently in social environments.

While chimps emulate, imitating the gist of an observed sequence of actions, infants over-imitate, encoding the specific series of actions they observe. Lyons’ study sought to understand why this difference between humans and primates exists, he said.

The answers may lie in our divergent evolutionary histories, Lyons said.

“The working idea is that we, as a species, have evolutionarily used more complex tools than our primate relatives,” he said. “As soon as humans started using tools recursively, how they worked became a lot more opaque. So it becomes advantageous for children to not make assumptions about tools, but to engage in high-fidelity copying.”

While Horner and Whiten’s finding may appear to show that monkeys have a more evolved learning style than that of humans, the opposite is actually true, said Andrew Young, a graduate student at the University of Wisconsin and the second author of the study.

In fact, Young said, children’s over-imitation tendencies help them learn more efficiently about new objects in their environment.

“Imitating others so faithfully has a rational adaptive basis,” he said. “The dominant idea is not that children mechanically encode needless information, but that over-imitation is a general learning mechanism that allows them to learn from social encounters more effectively.”

Infants’ ability to absorb involved details and extract a theory from these observations suggests that humans have an extra layer of cognitive complexity not present in primates, Lyons said.

But over-imitation is not always adaptive, he said — in some rare situations “when imitation may not be the best way to learn,” this tendency can work against children.

“[Sometimes] children’s ability to imitate can actually lead to confusion when they see an adult doing something in a disorganized or inefficient way,” he said.

But Lyons said the research is still too unclear to extend these findings into other domains, such as language acquisition.

He said he hopes to conduct future research to further probe the mechanism of over-imitation. The group is currently testing variable conditions to determine how long learned beliefs about a foreign object’s mechanics persist, when and how children stop over-imitating, and whether competition to open the object reduces the incidence of over-imitation, Lyons said.