Children learn more like scientists than adults. Children are "high-temperature" learners who explore wild possibilities with flatter priors, while adults are "low-temperature" learners who exploit known solutions with peaked priors. This makes children better at overturning established beliefs.
Children are oftentimes better scientists than adults.
Alison Gopnik, a psychologist and philosopher studying how children construct theories of the world, argued that children actively construct theories of the world, much as researchers do. They observe patterns, form hypotheses, and update their beliefs when evidence contradicts expectations.
The difference is in what Gopnik calls "priors." Adults have "peaked priors," meaning their extensive experience makes them stubborn. They've seen a lot, so they require massive evidence to overturn an established belief. When they change views, they do so incrementally.
Children have "flatter priors." They haven't accumulated as much experience, so they're more willing to accept unusual outcomes. They're not yet anchored to a particular worldview. This makes them better learners in some ways, more open to data that would surprise an expert.
Gopnik uses the concept of "simulated annealing" to explain this. Children represent a "high-temperature" phase: noisy, bouncy, random. They explore wild possibilities without worrying about efficiency. Adults are in a "low-temperature" phase: making small adjustments to existing knowledge, filling in details.
This is similar to the difficulty of training one's thoughts lies not so much in developing new ideas but in escaping the old, and reminds me of the quote "You are only as young as the last time you changed your mind".