"Son, one of the few things I've learned is that humans hardly ever learn from the experience of others. They learn - when they do, which isn't often -- on their own, the hard way."
-- Robert Heinlein: Time Enough for Love (p.38)
02Jan1995: What's the cognitive process which gives us the power of recursion over our environment? Is it problem solving? Is it perception? Analogy, maybe? Language use? Motor coordination? No, all of these are great tools, but I'm after something much more basic than any of these, a more minute building block in the puzzle of cognition. That process is learning.
The process which allow us to recognize and reconstruct the dynamic structure which we see all around us, the mental pipeline which forms abstract shells in which it fits other abstract shells or observed pieces of the environment, in a wonderfully recursive manner (with the stop clause[s] in the environment).
The only Computer Science prof from whom I actually learned something, Ronald Salley, used to say that everything in the world is recursion. Well, until last night I did think that's true. But a dream which I managed to catch in notes for a possible sci-fi story, made me consider the following: what if that's just another story coming out of our minds? What if recursion isn't actually in nature, but in our heads. Because in order for us to fit observations into our abstract patters and mathematical models, we constantly make approximations. What if recursion is simply our way to deal with the massive amounts of data which the environment throws at us every milisecond of our lives (or whatever the threshold of observation is for each of our senses)?
So what if I put together a neural network, like the one I wrote for that shape-recognition project, but I allow it to store its results after each training into new nodes which may have direct relations with the input nodes? Next time the same set of input nodes would generate a direct result, without having to go through training again. It would work like an oratoric memory device... Yeah, like a mnemonic device. So I'll call it a MneMonic net!
10Mar1998: That was what I have rushed down on paper sooo long ago (or so it seems), in a Volga room in AUBG after I had recorded the general lines of the dream from which I had just awoken and which was still buzzing in my head. Of course, over the time I polished the note from its original format, I wrote a BA thesis on the topic, and I kept chipping away at the project in my spare time until, in frustration (both with the model and with my personal life), I wrote a poem about it and a piece of confusing commentary.
The interesting part is that since I started studying Cognitive Science, I never had the time to work on it; there was always something else to read or to write, or to code, or my life seemed to rock out of joint, or I was working on other people's projects, then Cman came along and engulfed any spare time I might have, with joy and tears.
Now, if I manage to keep my position at Carleton and at the same time work for a living, I MAY have a chance at it as a dissertation project. If not, well, there are other things in life that are worth more than learning, and I already have two of them: a wife and a baby.
The worst part about learning is the realization that it gets to be somehow inconsequential. When you reach the level of specialization in a field that dictionary definitions start to seem incorrect, misleading or incomplete, you also remember that everything we know is subject to change and that we can know anything only of these things are constant or change in the ways we know they do. Our own world view matches with that of my neighbor only so much and even less with that of someone living in a different culture. And that's the point when I start to grudgingly appreciate the leveling effect of transnational companies like McDonalds or CocaCola. Yikes!
If you may want to see what was my previous step which led me to learning as a worthwhile pastime, you can consider the projects on my drawing board, or check some of my other pages available now: Literature, Computer Science and Cognitive Science. .