Predicting this post--and other plans
I predict, therefore I do. Or am. Or something
Instead of deciding to write this, I predicted that I would. I later revised my prediction. And now it’s been written. I predict I’ll post it in the next few minutes. If you see it, then let’s hear it for predictive processing.
Predictive Processing (PP) is a new theory of how the brain works. It argues that a brain is a prediction machine, not a plan executing machine. The brain makes predictions about the world based on a world model and then compares its predictions with sense data. Sometimes the sense data is adjusted to match the prediction—so we see what we expect. Other times the model is adjusted to reflect reality.
In the old model of brain function, the eyes collect data and send it to the visual cortex for processing. The visual cortex determined the scene’s contents and informed the rest of the brain. In the PP model, the visual cortex predicts what the eye will see and sends a signal forward to the lateral geniculate nucleus (LGN). The retina encodes visual data and sends it back to the LGN, which compares the sense data with the prediction. If they match, well and good. The LGN sends a difference signal to the visual cortex to update its model if they don’t match. (It’s a bit more complex because there are multiple such layers, and sometimes the sense data is adjusted to match the expectations.)
According to PP, the high-level functions of the brain control the body by predicting the position of parts of the body, and the lower-level functions activate muscles to make the predictions come true.
I read about PP in Scott Alexander’s SlateStarCodex blog in this post and this one and decided to apply the principle to my problems getting things done.
I remembered an earlier time, working with Elsa, my personal coach. I’d made a plan and didn’t do what I had planned. When we talked about it, I said I wasn’t surprised that my plan had failed. I’d made plans that had failed hundreds of times before. My failure was unsurprising. We then worked to make a plan that might fail—but whose failure would at least be surprising.
Said differently: After making my first plan, I would have predicted failure if I had asked myself to predict the result. Why? Because most of my plans fail. Most don’t even deserve to be called plans. Vague intentions are more like it. So I needed to make a plan that I would predict would succeed.
Psychological research tells us we are poor predictors of our behavior. So after deciding to do something, I’ll ask myself: do I predict I’ll do it? If I don’t predict success, I must decide whether to do it. If I do, I need to predict success, at least.
But there’s an even better question: would I be surprised if I failed to follow through? If I weren’t surprised, my prediction is probably too generous. But, on the other hand, if I make a decision that I care about, I need to predict that I’ll succeed and be surprised if I don’t.
I predict I will post this as soon as I finish this sentence and convert the post from markdown.
Update: 30 September 2022. Apparently, I predicted this might be useful for someone I was talking with on a Discord server. I predicted I would edit it because I did. Grammarly found many errors and helped me improve my writing, which I probably predicted. And now, I predict I will update my post and share a link with the person who inspired these predictions.
Update 2: I did not predict two misspellings in the last para, but a friend saw one and predicted I would correct it, and he was right.