Praising learning accidents

October 8, 2023

I have always been an unremarkable football player. I occasionally joined some informal games in the breaks between classes. More when I was younger, and less often as I grew older. By middle school (grades 7-9) I barely played. I was much more successful at miniature sports: ping pong or table soccer.

However I remember this one afternoon in 8th grade, when I joined a game by the insistence of my classmates. I thought it was all pretty unremarkable, until after the game, one of my friends came to me and praised my passing “technique”. He had noticed that I tended to kick the ball in an arc, lifted above the ground, when trying to pass to another teammate. And this, he said, made it hard for the other team to intercept.

The thing is, this was purely accidental. I was not trying to make tricky passes, so much as I was trying to get rid of the ball without stumbling on it. I was overloaded enough with the basics. But here was my friend, who played football competitively, praising my awkwardness as if it were a skill!

It made me wonder, could I actually be better at this than I thought? In the end, my “trick”, which I repeated a couple of times in future games (now consciously), was not enough to make me a good player. But I never forgot that comment! It is a memory I still hold dear.

I was reminded of this experience when reading the colorful “Surely You’re Joking, Mr. Feynman!”. There’s a story surrounding Feynman’s experience of learning how to draw as an adult:

Jerry turned out to be a very good teacher. He told me first to go home and draw anything. So I tried to draw a shoe; then I tried to draw a flower in a pot. It was a mess! The next time we met I showed him my attempts: “Oh, look!” he said. “You see, around in back here, the line of the flower pot doesn’t touch the leaf.” (I had meant the line to come up to the leaf.) “That’s very good. It’s a way of showing depth. That’s very clever of you.”

“And the fact that you don’t make all the lines the same thickness (which I didn’t mean to do) is good. A drawing with all the lines the same thickness is dull.” It continued like that: Everything that I thought was a mistake, he used to teach me something in a positive way. He never said it was wrong; he never put me down. So I kept on trying, and I gradually got a little bit better, but I was never satisfied.

Here too, is a competent person praising the technique of an amateur, who is engaging with the task at a much lower level of competence and ambition. Neither I nor Feynman were foolish enough to believe these were more than “accidents”, and yet the mere fact that someone we respected praised them genuinely, turned out to be immensely rewarding (Feynman went out to become a much better drawer than I ever became a football player). These outcomes happened by chance, but they were also easy enough to repeat now that someone had brought our attention to it. So we could already do something interesting! And, in a sense, they were achieved for free. What other accidental techniques may we find next if we only keep practicing?

The most common strategy to teach animals how to perform complex tricks is called “shaping”. It boils down to rewarding spontaneous small actions that seem necessary to perform the trick as a whole. For instance, you can easily teach a chicken to press a lightbulb switch by rewarding it (giving it food) everytime it makes a step closer to the switch, and once it gets there reliably, everytime it pecks on the switch. In a matter of minutes, you can get the chicken to systematically move towards the switch and flip it on! This works with all sorts of animals… and perhaps humans too.

The trainer must wait for the animal to act spontaneously and reward those actions that seem the most promising (or surprising). Because they come unforced, these actions have an interesting mix of individual proclivity and randomness. They could be the seeds of a unique personal technique later down the road.