If you flip a light switch and nothing happens, there are a couple of possible explanations. One is that something has gone wrong in the external world — maybe the bulb has burned out. Alternatively, you may have made a mistake, perhaps flipping the wrong switch.

Infants can integrate prior knowledge with statistical data to make these distinctions at a very young age [...] 16-month-old infants can, based on very little information, make accurate judgments of whether a failed action is due their own mistake or to circumstances beyond their control.

[...] very young infants can quickly learn basic principles about how the world works, then use those rules to interpret the statistical evidence they see. [...]  “They can use very, very sparse evidence because they have these rich prior beliefs and they can use that to make quite sophisticated, quite accurate inferences about the world.”

In one condition, babies saw a toy that played music when one experimenter pushed a button on the toy but failed when a second experimenter tried, suggesting that the failure was due to the agent. In another condition, the button sometimes activated the toy and sometimes failed for each of the two experimenters, suggesting that something was wrong with the toy.

[...]

Depending on the circumstances of the experiment, the babies did respond differently, indicating that they were able to weigh evidence for each explanation and react accordingly. Infants who saw evidence suggesting the agent had failed tried to hand the toy to their parents for help, suggesting the babies assumed the failure was their own fault. Conversely, babies who saw evidence suggesting that the toy was broken were more likely to reach for a new toy (a red one that was always within reach).

[...] 16-month-olds could use very limited evidence (the distribution of outcomes across the experimenters’ actions) to infer the source of failure and decide whether to ask for help or seek another toy. That finding lends strong support to the probabilistic inferential learning model.

“What the finding here seems to be is kids are smart, but the way in which they’re smart is really tracking statistics. They’re really good learners,” she says.

Link: web.mit.edu/newsoffice/2011/assigning-fault-0624.html

PDF: web.mit.edu/~hyora/www/Hyo/Research_files/GweonSchulzCogSci2010.pdf

Study by MIT cognitive scientists Laura Schulz and Hyowon Gweon appears in the June 24 issue of Science