Hi guys! I hope you all had a lovely Christmas or whichever solstice festival you celebrate, and a great New Year too. I've been away, so I don't have much to report. I did drive all the way to the east coast and back again, though, so I had plenty of time for thinking. I just transcribed my notes from iphone voice memos to text and I'm sure they must have meant something to me at the time I made them… If you're of a truly masochistic bent, you can read them here, but I doubt you'll have any more idea what I was talking about than I do, and that's not much! 😉
For the techies, I've basically been musing about the fine details of the Yin learning mechanism in the creatures' brains and arguing with myself about the merits of two different approaches. One of them is a just-in-time method, in which neurons recognize the moment when they are well placed (literally) to learn something, and the other has them learning pretty much at random and then those memories migrating across the cortical surface to the best locations. One of my basic hypotheses is that perception, intention, attention, speculation, thinking and planning are just different ways of looking at essentially the same thing. On the upside this is a wonderful buy-one-get-one-free deal, because all these apparently different things come out of a single mechanism. On the downside, though, it means when I'm trying to flatten down the bubbles in the wallpaper at one end, I have to make sure this doesn't make new bubbles show up somewhere else. Any decision I make while thinking about perception and categorization has knock-on effects in the areas of intention and attention that I have to think through to some extent so as not to paint myself into a corner (to stick to the decorating metaphors), so weighing up these two options is tricky. Today I'm coming down on the side of migration. Tomorrow it'll be the just-in-time approach…
But on the whole it's all coming together quite nicely. Just before I left for the east coast I found a solution to a problem that's too tedious to explain but turns out to fit quite nicely with some other things. Designing brains is a weird business, but you get these fantastic moments when you solve one problem only to discover it also solves a bunch of others, and then you just know in your heart that what is emerging in front of you has some truth behind it. Unfortunately there are also moments when you solve one problem only to create three others, but I try not to dwell on those!
Sadly there's still nothing to see – I have a brain scanner that shows some pretty colors as the different layers of neurons fire, but frankly, who cares? When is the damn creature going to actually DO something? Not very long now, I think, and thanks for being so patient. From a neuroscience perspective, these colored patterns are like the first glimmers of life – like someone starting to come back from a coma. But from your perspective it's mostly behavior that counts, and until I get two complementary brain systems functioning there isn't going to be any behavior at all.
I'm started to get excited, though. I remember the first moment a norn actually made some decisions under its own steam and it's really quite freaky to see something you've made get up and walk because it felt like it! There'll be a first moment like that here, too, and I'll be sure to share it with you. The behavior will still be barely distinguishable from random, so don't expect Gollum to look up and start quoting Descartes! But it'll be great when the major pieces are all in place and start working together in some semblance of harmony. I don't think that moment is too far away now.
Assuming, of course, that I can figure out what my notes actually mean…
Hope you're all well!
– Steve
Leave a Reply