Friday, March 30, 2012

Pattern Recognition Bug Fix

Mainly because I'd prefer to write posts that can lead to concrete action. It's not like philosophy is any more incapable of leading to engineering than any other field of inquiry.

Secondly, to promote communication. Maybe this trick is so well known nobody else thinks it is worth talking about. Or maybe some opposite.

This kind of thing is what Lifehacker would have to write about for me not to be disappointed in it. It would be great, as it would likely do the writing part way better than I do.

Inevitably this lead into laying out some of my thoughts on how the brain functions at less abstract levels.

There's a well-known bug in the human brain's pattern recognition modules, in that it is too aggressive. It flags many patterns on insufficient evidence, and often sticks to them despite additional evidence.

One day I wondered what would happen if I set my pattern recognition modules to work at recognizing the patterns characteristic of spurious patterns. I actually felt my brain go clunk. (Almost immediately? Next time I did pattern recognition?) Thinking became easier and more comfortable, and I found out I'd been fighting myself due to sudden onset peace syndrome.

Two and a half issues. First, a clunk feeling is not normally considered good scientific evidence. First and a half, a feeling of ease - while an end into itself and therefore still worthwhile - is not normally considered good evidence of ease. Second, this immediate reaction may have been based on certain other habits I'd developed, which anyone wanting to replicate this may not share.

I'm a bit socially stupid. When someone told me to question everything, I took it seriously. When I questioned whether the category that 'clunk' belongs to functions as good evidence, I found the scientists were wrong. I cannot find any strong verification of their position - I can only re-derive it from certain historical accidents. (Suggestions are welcome.)

For example, when I tested this clunk by checking my new-feeling pattern of patterns for spurious patterns, using the exact same methods I'd used before, I found spurious patterns had basically disappeared. (I still spot check sometimes.) The simplest explanation for feeling better is that it is better, and in this as many other cases, this was borne out.

Secondarily, human brains are far more similar than different. The clunk feeling should be repeatable, even if it is misleading. So, don't worry about whether its misleading, first worry about replication. Once replicated you can just test whether its reliable, rather than having to argue about it. In any case, a difference in feeling has to reflect a difference in function, the only questions are what and how much.

As for my habits, I had been deliberately training myself in inter-brain communication, as I had been raised to do. My pattern recognition modules can hear and understand my consciousness module, and so all I had to do was consider the idea and it became done. Whether this is normal or due to the training, I don't know.

If it isn't normal, getting the pattern recognition module to hear and understand the idea may take some finesse. The subconscious is suggestible, which means it can hear and often understand the audio-processing modules, which means saying it out loud may be enough. However, it is possible for the subconscious to hear, understand, but not to agree, in which case it won't obey. I can't imagine how a subconscious not agreeing with this, but my imagination is limited.

Another pathway is to practice it deliberately. The term is 'automaticity' but what's happening is the subconscious is understanding and agreeing with the procedure, and takes it over. The subconscious has to handle all the fiddly little details like muscle nerve signal encodings and information storage, retrieval, filtering and processing. You can say to it, "Find the patterns characteristic of spurious pattern" and it will just reply, "Okay, but what changes does that mean I should apply to these filtering and processing algorithms?" If you can't explain, you can lead by example.

And for your own sake, if it pushes back, listen. (In me this even feels like backpressure.) Your brain is on your side. Its interests cannot be more aligned with yours. Your consciousness directly commands something less than 0.1% of your neurons, and my estimate is inflated because consciousness is computationally inefficient. Much of the rest of the brain isn't general purpose, but the sheer weight isn't just for show.

No comments: