Start by disqualifying the question. Nobody has taken a group of conspicuously independent adults, traced them back to childhood and catalogued the lessons their parents delivered. That study does not exist, and any list presented as though it does is a list of guesses.
What does exist is narrower, better and stranger. There is a randomized experiment on what a single sentence from an adult does to a four-year-old’s willingness to investigate, and a study of how six-year-olds spend their weeks that relates that time to one careful measure of doing things nobody told you to do. Between them they suggest that the thing which produces independent behavior is not a lesson at all. It is a specific and expensive absence.
Eleven words and a toy
Elizabeth Bonawitz, Patrick Shafto and colleagues built a toy out of colored PVC pipes with four non-obvious things it could do. Pulling a yellow tube out of a purple one made a squeak. Elsewhere on the board there was a light, some music and a hidden mirror. Then they gave it to 85 preschoolers in a science museum, one at a time, in four conditions.
The Artful Age
A weekly letter on aging well, family across generations, and the creative life after the kids leave home.
In the teaching condition the adult said: “Look at my toy. This is my toy. I’m going to show you how my toy works. Watch this.” Then she pulled the tube, made the squeak, and did it again.
In a second condition she said exactly the same thing and made the same squeak, then interrupted herself and hurried off, as though she had more to show and ran out of time. In a third she pulled the tube as if by accident, looked surprised, and said, “Huh, did you see that?” In the fourth she showed the child nothing at all.
Every child then heard the same line. “I’m going to let you play and see if you can figure out how this toy works.” And every child was left alone with a four-function toy.
The taught children performed an average of four distinct kinds of action on it. The interrupted, accidental and untaught children performed 5.3, 5.9 and 6.15. The taught children spent 68 percent of their play on the one function they had been shown, against 53 and 38 percent for the children who had seen the same squeak produced in the other two ways. They quit sooner, after about two minutes against nearly three and a half for the children shown nothing.
And they found less. Scored out of the three functions nobody had demonstrated, the taught children discovered 0.72 of them. The interrupted children found 1.3, the accidental 1.2, the untaught 1.15.
The children were right
Here is the part that turns this from a scolding into something useful. The authors do not think the taught children were incurious or passive. They think they were reasoning correctly.
If a knowledgeable, helpful adult sets out to show you how her toy works, and shows you one thing, then the sensible inference is that there is one thing. A teacher who knew about the light and the music and the mirror would have demonstrated them, because that is what teaching means. So the absence of evidence for the other functions is, in that context, real evidence of their absence. The sentence “I’m going to show you how this works” is not only an offer of information. It is a quiet claim to have covered the subject, and preschoolers hear the claim.
The second experiment is the one that seals it. Children narrowed their exploration in the same way after merely overhearing the lesson delivered to another child. They did not narrow it when the identical lesson was given to an adult, in most cases their own parent, or when the adult performed the action with no teaching frame at all. Four-year-olds were tracking who the lesson was aimed at, and treating a lesson for someone like them as a lesson for them.
The week, not the moment
The other study works at a completely different scale. Jane Barker, Yuko Munakata and colleagues at Colorado asked the parents of 70 six-year-olds to reconstruct the previous week in a grid of half-hour cells, writing not “free time outside” but “played tag outside with older sister and friends from the neighborhood”. Three raters, blind to everything else, sorted the entries into structured and less-structured time.
The children were then tested on two different kinds of executive function. One was the externally driven sort, following rules an adult has supplied. The other was self-directed: generating your own approach with nobody specifying it.
More less-structured time predicted better self-directed control, holding age, vocabulary, gender and household income constant. More structured time predicted worse, although that result was only marginal and deserves to be reported as such, since popular accounts of this study tend to promote it. Neither kind of time predicted the externally driven measures at all. Whatever is going on, it is specific to the business of directing yourself.
What the researchers will not say
Neither team draws the conclusion the genre would draw. Bonawitz and colleagues write that there is no fully general answer to whether the costs of instruction outweigh its benefits, that it depends on how much the teacher knows and how likely the child is to find things alone, and that teaching does what it is supposed to do, which is produce fast and efficient learning. The remedy they actually suggest is a small one: delay the instruction until the learner has had a chance to investigate first.
The Colorado results carry heavier caveats, and the authors list them themselves. The sample was mostly affluent and suburban. Self-directed control was measured with a single task. The time diaries came from parents rather than observers. Their categories are admittedly imprecise, since a library visit and a sporting event were both filed as less-structured. They ran multiple exploratory analyses afterwards and say plainly that these are not ideal conditions for statistical inference. The design is correlational, and they raise the possibility that the arrow runs backwards, with children who already have weaker self-direction gaining the most from open time. There was also an oddity they did not predict: the relationship held for the younger children in their narrow six-to-seven-year band and vanished in the oldest quartile.
The two papers do not cite each other. They arrive at a similar-sounding idea from opposite directions, which is worth something, but agreement is not the same as proof, and neither one tested the mechanism the other implies.
What is actually left
Not a list of counter-intuitive lessons, because the evidence does not contain one. What it contains is a trade, and the trade is legible.
Every time an adult says “let me show you how this works”, the child gets the answer quickly and stops looking for the others. That is a good bargain when the adult knows everything worth knowing and the child would not have found it alone. It is a bad bargain when the thing has four functions and the adult knows one. Most of life is closer to the second case than we like to admit, which is the whole argument for ordering it differently: let them have a go, then show them.
And the smaller finding underneath it is the one worth keeping. The children did not need to be told to stop exploring. They worked it out from the way an adult introduced a toy. Whatever independent people were given, some of it was probably given by accident, in the gaps where nobody announced that they were about to explain.