Hacker News Reader

Comment by ubertoop | Hacker News Reader

I'm trying to read this, but a lot of it is way over my head. Would you mind explaining it?

The blog he linked, Slate Star Codex, has a lot of good posts on predictive processing theory. I'd recommend checking them out. (The author is a psychiatrist, and in my opinion a semi-polymath.) Basically a theory positing that the brain fundamentally works, at least in part, by continuously updating a model of the environment and the self based on successful and unsuccessful predictions.

Although this may not seem like a crazy idea, some proponents suggest that essentially everything the brain does works in this way. For example, that the cause of schizophrenic symptoms may be due to them holding a very loose predictive model with weak priors; basically that they don't have pre-existing high-confidence predictions about anything, on every level. You and me likely would predict with very high confidence that we are not and can't be a god, but someone with schizophrenia may not predict that with as much certainty due to a general issue with neural prediction.

I am oversimplifying it and may be partially wrong, but I think that's the general. I'd highly recommend reading that blog's series on predictive processing for a much better explanation. The linked post also discusses schizophrenia.

Thank you! I appreciate the summary. That actually helped a lot... I'm reading through some of the articles on this website, but feel like I'm lacking some prerequisite knowledge. I find this stuff very interesting just, a bit out of my league. Must keep reading I suppose :)

If this theory is correct then one could use it to guide one's own life?

Here's something I don't understand when trying to look at it through the lens of this theory:

I've had mostly success when it comes to software development. Whether it be new languages, new projects, what have you... If I sit down, focus, and do the work, it comes out well and succeeds. This is in stark contrast to other areas of my life btw, where I've worked very hard at things but made minimal progress.

So if I understand correctly, my internal prediction machine should be predicting success, and thereby providing a jolt of motivation, whenever I think about a new cool project.

Take VR development. I really want to learn Unity and make a cool VR game. But every time I start, I putter out. I get distracted, I don't make progress fast enough, I just.... end up stopping, and doing something else. Why?

Because in a new domain, you don't know which pieces to tug at to lead to success and which not to. When you try something and fail, you don't retrospectively identify which parts of your attempt led to that failure. So instead of a clear nuanced picture like "do X.1 never X.2...sometimes X.3 simultaneously with X.4", you've just trained yourself to predict "If I do X, I'll likely fail."

You need something to point out which bits of X to pay attention to. Thats why teachers are valuable -- the indicate where to pay attention.

---------

Consider going onto http://codementor.io/ and hiring someone to pair-program with you for a while. If you were at the office and spent 4 hours getting nowhere with a problem, you'd ask a colleague for help.

Reply on news.ycombinator.com