Up until now, metacognition (the ability to reflect on one's own thoughts and actions) was considered to be almost non-existent in children under the age of 6 or 7. But a recent study has revealed that, beginning at 20 months, babies are capable of (non-verbally) expressing their own uncertainties. How did researchers manage to overcome the poor language abilities of young children in order to demonstrate this reflective ability in very young children?

Metacognition allows us to optimally acquire new information by adapting our learning strategies according to our current knowledge state. Because of this, metacognition has proven to be a reliable predictor of learning. In pre-school children, evaluating metacognitive abilities requires developing a protocol that doesn't require language.

In the study carried out by the Brain and Consciousness team at the laboratory of cognitive science and psycholinguistics (CNRS/ENS/EHESS) and published in the medical review PNAS (Proceedings of the National Academy of Sciences), the French researchers recount the following experiment. They had 80 babies (average age = 20.17 months) memorize the location of a game hidden from their view by one of two boxes displayed in front of them. After varying amounts of time during which the boxes were placed behind a black curtain (for 3, 6, 9 or 12 seconds), the babies were supposed to point to the box where they believed the toy to be. But there was a fifth level of difficulty in which the boxes were directly hidden behind an opaque curtain, making it impossible to know where the toy was hidden. The goal of these manipulations was to determine whether the infants could monitor and communicate their own uncertainty (particularly for long waiting periods and the “impossible” condition).

What did the researchers find? Compared to the control group, in which the babies had no choice but to decide for themselves, the children in the experimental group, who had the opportunity to ask for help using non-verbal communication (when they had forgotten the location of the toy), strategically used the option to point to the right box. This shows that infants know when they don't know, can show it, and can share this information in order to reach a goal. In addition, the difficulty of the task had an impact on the probability of asking for help. Babies were more likely to ask for help on the “impossible” task than on the “possible” ones, and for the latter, the likelihood increased with memorization times.

Another interesting result: babies generalized their help requests even for “possible” tests in order to improve performance. Finally, the fact that children in the control group didn't spontaneously ask for help when they were unsure shows that they needed to be made aware of this possibility. The fact remains that 35% of children in the experimental group didn't take advantage of the help. This behavioral difference reflects differences in metacognitive abilities, particularly that some children tend to overestimate their abilities.

Other factors such as executive function and parental attachment (the parent being the helper in this experiment) may also play a role in the interindividual differences in metacognitive abilities and help-seeking behavior seen in the experiment.

In conclusion, the researchers suggest that their findings highlight the usefulness of explicit metacognition not only for cooperation, but also for learning from others.