|
We have a bunch of hardwired algorithms, and a lot we have to learn in our lifetime. Let's say we have a "main" that when your brain gets the hunger signal you want to eat (or if you're a baby you just cry because you can't fetch your own food), that wants to sleep, wants to self-accomplish, find a suitable mate, crave social acceptance, etc. Our learning algorithms also have very basic buiding blocks. We deduce that things that are close together in time or space have something to do with each-other. If I press a switch the light goes on. That's how we learn causal effects.
A lot of things are partially hardwired. Like language, we have structures in place for understanding grammar but we need to learn the words. Another example, we know hardcoded what emotions mean, but we need to learn how to express them by mimicking our parents.
Other things are fully hardwired. Like our perception, which most of my class was about. Interesting is for example the way we perceive motion. Basically your brain stores images of previous vision (snapshots) and runs them through an AND gate with what you are seeing now to find out what is different. And we do this comparison at different frequencies, for differing speeds. That is why we can't see really fast motion, but also, and most don't know this one, we also can't see really slow motion. (that is why the bad guy goes for his gun slowly, trying to not make you see it by going at a speed slower than your slowest comparison frequency)
Ok going off on a bit of a tangent here, but just to show how algorithmic a lot of the stuff we do is, even though we're not aware of it. And sorry if this sounds preachy, i'm a technical guy :P
Originally Posted by a500lbgorilla
While I'm sure you could write an algorithm that learns in this manner, it would only show that algorithms can approximate and simulate the brain, not that the brain runs algorithms.
Why is it only simulate? After all the brain is just electrical pulses, nothing too outlandish or fancy there. What's the difference?
|