An amazing feat of human cognition was recently recorded at MIT. What gives this extra kick is that the subject is still in diapers.
In an article published in the May 2011 issue of Science, Hyowon Gweon, a graduate student studying childhood cognition at MIT, showed that 16-month-olds are able to use logic—a skill, it is worth remembering, that nobody has taught them at this age—to infer the cause of failed actions. The study shows that 16-month-olds can use a very small amount of information to reach important conclusions about why an action has failed.
The researchers gave each baby a toy with a button on top and showed that by activating the button, the researchers could make the toy play music. When the babies pressed the button themselves, music did not play. This created a quandary for the babies: did the cause of the failure lie in themselves or the toy? As Gweon points out, this is a kind of problem we encounter often in adult life, as when someone has trouble printing a document from his computer. “Is it because the computer’s not working,” she said, “or is it because you’ve done something wrong?”
The researchers were able to show that the babies used the information they had learned from seeing the first demonstration of the toy and apply it to the new situation, concluding that the failure was the fault of the toy. It might not seem like much, but it is a fairly impressive display of reasoning for sixteen months. From just small amount of data, infants can put together larger connections to attribute cause to effect.
This ability to do much with small inputs is what makes human learning such an extraordinary and mysterious process. Gweon’s advisor, MIT Associate Professor Laura Schulz, says that many of the greatest feats of human ingenuity happen in the first few years of life.
“If you take all the hard problems of cognitive science,” she said, “of face recognition, spatial navigation, number learning, learning a natural language, learning moral reasoning, learning causal relationships, learning about other people’s roles and intentions and mental states—you learn all of that within the first few years of life. So by the time you’re three, you’ve mastered all of those areas. You’ve mastered them in an unsupervised way—no one has explicitly taught you these, no one has rewarded you for learning common-sense knowledge about the world. And every typically developing child has all of that by the time they’re three or four years old. It’s astonishingly rapid. We don’t know any other organism in the world that can do that, let alone any artificial intelligence system that we can make do anything like that.”
Fascinated by the brain since childhood, Gweon initially thought that she would study human memory. In her senior year of college, however, a lecture by a visiting Yale professor convinced her that childhood development offered even more exciting questions in the field. At MIT, she works in the Early Childhood Cognition Lab, where she says an atmosphere of intellectual cross-fertilization and debate, fostered by the close proximity of researchers in the “hard” and “soft” sciences, helps to inspire exploration and creativity.
That creativity came into play when it came to designing the experiment. In the sciences, asking questions is easy, but designing experiments that address them accurately is both a skill and an art. This may be especially true in the field of child cognition: researchers first have to translate their questions into a standard structure, with two hypotheses and two competing arguments, and then they have to translate their premise into child-friendly form—while making sure the inputs the children receive remain unambiguous. Gweon and Dr. Schulz started with a clear question—can infants use logic to tell the causes of certain actions?—but when they brainstormed possible experiments, none seemed workable. Finally, Gweon proposed an experiment involving manipulating a toy. Dr. Schulz initially protested the setup as too complicated, but Gweon was able to persuade her it would work. “I said, ‘If it does work, it’s exactly the right study, but there’s no way it’s going to work,’” Dr. Schulz said. “I lost that bet, and I’m very grateful for it.”