04/18/2010 By Shawn Burns

Teaching means being happy with, looking forward to, and deliberately designing situations intended to end in failure.

I make sure my kids fail often. There is no learning in success.

I am happy when Erin climbs a structure on the playground, but I recognize that her success there hasn’t really taught her anything. It’s when she falls a little, slips, dips, misses the ring, that she is really learning how the playground works.

The human brain is an insanely complex prediction machine. It predicts from early on, and the only way it changes, adapts, and grows is through failures of prediction.* When it predicts the world will be blue, given the inputs it has received to date, and the world turns out red, the network of interconnected neurons adjusts itself and fine-grains its distinctions between items that it previously took to be evidence of “blue world.” But if the world turns out to be blue the network doesn’t change at all. A tiny brain in a tiny body that never failed to predict outcomes would remain a tiny brain (metaphorically speaking).

Here someone might wonder “But what about all the “reinforcing good behaviour” stuff we learn about, that we’re taught to use in order to produce happy, well-adjusted children? Isn’t that teaching through success, rather than failure?” Well, no. Because it isn’t teaching the network how to predict the future; it is conditioning the network to perform in a certain way according to the wishes of the teacher. It may very well be the case that the best way to produce a well-conditioned network is through positive reinforcement rather than negative, but that network will not have learned how to predict the future any better than if another method had been used to condition it. Conditioning the network makes the network predictable, not the world.

Part of parenting is making our little bipedal networks predictable. But parenting doesn’t end there. If we want to teach our kids, to make them as psychic as human beings can be, as rational as human beings can be, as attentive to the fine distinctions between facts as human beings can be, we need to make them fail.

Traumatic failure is a great teacher, and it’s embedded in folksy wisdom like “Fool me once, shame on you. Fool me twice, shame on me,” where the point is that the trauma of being fooled ought to have been enough to teach you how to avoid being fooled in that way again. Similarly, it would be hard to find someone who, after having been burned by sticking his fingers in a flame, would go around doing the same thing again. Traumatic failure teaches sharp, indelible lessons.

But it can be an over-teacher. Rather than learning only as much as we should about the world from the experience, we might “learn” things about the world that aren’t true, but which will still be carved in our brains because of the trauma. For instance, burning a hand on a stove element might result in a fear (a prediction about the world) of stoves rather than the elements.

This is one reason why corporal punishment is both loved and hated by parents: it is a shortcut to the traumatic lessons the world might give, and so bears the hallmarks of strong natural lessons, but it is also an over-teacher: the lesson learned might not be that some action is wrong or dangerous, but that adults are inclined to hit you for doing it. 

Further, it resembles conditioning in the way it creates sometimes arbitrary, or teacher-directed, behaviours; insofar as it makes the brain less able to predict the future based on evidence it is completely counter-productive to teaching.

Teaching through failure doesn’t mean creating artificial traumas as a shortcut to learning how the world works. Teaching through failure means making the choice to allow safe, but vivid, lessons to take place. It means letting kids get dirty and to figure out how to not get dirty. It means focusing on the effort and method and not the outcome.

Who is disagreeing with this? You might wonder about the point of taking the time to write this down, when we all nod our heads at it: “Of course failure matters. Of course it’s a better teacher than success. I knew that already.” But someone out there is getting it wrong. Someone is trying to teach through success, focusing on empty accomplishments rather than training brains to winnow the world. Someone is happy with the outcome if a functional illiterate gets straight A’s in college classes and is given a job making stupid choices, but gets paid well to do it. Someone is convinced that recognizing the accomplishment is more important than whether there was something to be accomplished at all.

I know this because I look at my own life and I see that I was never, ever put on the path to effort, only the path to success. Some natural abilities I have made getting the recognition for accomplishment easy, even if there was no work involved for me. And I don’t think I’m alone. I think a generation of kids has been taught the same lesson. We think, for example, that it’s better to find the warp holes in Super Mario Bros. so the game can be finished quickly than it is to explore every level and see what the designers built. We didn’t learn anything except how to subscribe to magazines that sell cheat codes. Yoda taught us all a miserably bad lesson: Do or do not, there is no try. So we don’t try, for fear of not being able to do.

We learned how to get A’s on tests, and that it didn’t matter if we forgot everything we were tested on shortly after: the content of the lessons was never important, only the outcome.

There is no try? There is only try. Trying is the most important part of doing. Try, for try not and there is no do. Gretzky said you miss 100% of the shots you don’t take. Don’t wait for the goal to come to you, in other words, or for the stars to align and magically grant you success; take shots, fail, and take more shots. You will learn how to shoot, and you will learn what is involved in shooting. The world will become predictable for you.

This is all a long way of saying that yes, Emily, I did let Adrian eat a yogurt popsicle and get it everywhere and try to hold a paper cup of water and drink from it and really just pour water all over himself. It was all because of this philosophical commitment I have to teaching through failure and not, rather, that I thought it was funny.

I’ll do the laundry.


*This is not really an accurate statement. There is a kind of learning that happens in networks where the connections are strengthened just in virtue of them being strong all along. So the more often the network is excited by the world turning out the way it predicts, the more it will predict the world to be that way. If the world continues to be that way the network will have been appropriately adjusted, in retrospect. But it’s hard to see something like that as genuine learning, where learning tracks truth, rather than a practically beneficial coincidence. Some might claim that a practically beneficial coincidence that results in outcomes that are identical a results from a properly trained-up network is a distinction without a difference, that either both are learning or neither are. But that way lies a philosophical school called “Pragmatism”, and I’m not a fan.