On December 1 I wrote about
trading and the problem of random reinforcement, one of my more popular posts. But what happens in those areas of trading that don’t fall prey to the random reinforcement of the markets? That is, what happens when we are actually rewarded for doing something right and not rewarded or punished for doing something wrong?
A column by Scott Berinato in the latest Harvard Business Review looks at a study of monkey learning where reinforcement was not random: the monkey was rewarded if he did a certain task correctly and got no reward if he did it incorrectly. Importantly, the researchers expanded the literature by monitoring the brain activity of the monkey both during and after he performed the task.
Old saws are often true. In this case, that success breeds success. After the monkey performed the task correctly “neurons in the prefrontal cortex and striatum, where the brain tracks success and failure, sharpened their tuning.” Moreover, the changes lingered for several seconds, “making brain activity more efficient the next time the monkey did the task. Thereafter, each success was processed more efficiently. That is, the monkey had learned.”
After failure, however, there was little change in brain activity. The brain stored no information about what had gone wrong, even though the task was simple in structure: if A, then x; if B, then y. There weren’t exactly 1000 ways to fail. But the brain didn’t know what to store, so the monkey didn’t learn from his failure. “The monkey just tried, tried again.”
Indeed, this is the problem with failing. In most cases if we knew why we failed, we probably wouldn’t have failed in the first place. If we take a second stab at the problem without knowing what went wrong the first time around, assuming that we are bright (or sane) enough not to do exactly the same failing thing again, we probably won’t be any better off. Think about trying to solve the Rubik’s cube without a game plan or an online cheat sheet. It’s just the poor failing monkey at work again.
So if, to make this research relevant to everyday trading decisions, we’re confronted with the problem of random reinforcement in our individual trades and if we can’t make sense of many of our non-random failures, the only thing we can hang onto are our non-random successes. Well, that’s not the end of the world. We all have some successes—from clicking on the right button to enter a trade to controlling risk appropriately. And success (with practice) breeds success, though of course it never manages to eliminate non-random failures.
Also, on the bright side, since we aren’t monkeys we can read about how to do things. Want to install a lock and haven’t a clue how? The Internet will provide instructions. Want to build a trading system and don’t know where to start? Again, there are lots of resources, from books to web sites to blog posts! So we don’t always have to be on the frontier; we don’t have to be Edisons discovering a thousand ways not to do something. We’ll still have plenty of non-random failures, many of which we won’t learn from. But between successes and instructions to prevent failure we’re a lot better off than the laboratory monkey.