[Retric]

Your thinking of a basic classifier Neural net from an AI class. But, that's a poor analogy for anything more complex than reflexes. We have both short and long term memory, and we can do planning. A really simple way to thank about short term memory is to have a single neuron in a feed back loop. Basically it's output leads to it's input and it can be switched on and off. Of course real neurons have far more than just 3 connections and tend to work in fairly large networks, but it's gives you an idea of how you can have a short term memory you keep cycling though the lyrics of some song and it's 'stuck in your head'.

Long term memory is a physical change in the layout of a neural network. Spend long enough walking around a new city and there is an abstract but physical map that's actually stored in the layout of neurons in your head. At the same time pieces of that layout represent locations on the map and how they connect to other locations. Think of an entrance to a parking lot and you might have a fairly static picture of the location linked to the choice of where that will take you if you go there. (So short term memory is now a neuron that cycles, a network that picks which target network to activate, and a network that represents something, plus feedback to turn on and off the targeting.)

So what's consciousness? It's the ability to think about things as abstractions. When you say Apple to your self your actually activating neurons that keep cycling Apple over and over. Picture yourself tossing a ball and your thinking thinking about starting the cascade of muscle memory that causes you to through something while picking the perimeters of where you want the ball to end up and how hard you want to to hit the target, and possibly the position you need to be in to actually be able to through the ball. You can play around with the outcomes of if I do this that will happen by checking what neural net's predict the outcome will be. Chess is a great analogy for this, players don't think about moves as picking a piece up and moving it somewhere else, but what it means when the piece is in a new location. Though experience, education, or just thinking about things you can even train these neural networks to get better at those predictions.

Of course the actual implementation of these things is horribly complex, and many of the specifics are not all completely understood / studied. Also, chemicals play a major role, there are actual chemicals that represent things like pleasure in the brain. EX: Cocaine mimics the mostly hard coded chemical reward response for things like having sex by blocking the dopamine reuptake transporters. http://serendip.brynmawr.edu/exchange/node/1704

PS: Still I hope this simplified model helps you understand what's going on.

[ma2rten]

You seem to be missing my point.

So what's consciousness? It's the ability to think about things as abstractions.

No, it's not! (see http://en.wikipedia.org/wiki/Consciousness). Artificial NNs can also model memory (see Hopfield networks). However, I just used them as an analogy. Maybe Zombies work better for you and the Philosophical Zombie Wikipedia page does a better job of explaining the idea (thanks again xyzzyz).

[Retric]

Perhaps if I reword that a little you might understand what I mean: http://en.wikipedia.org/wiki/Consciousness is the subjective experience of the ability to think about things and processes as abstractions. Apple, Airplane, Alphabet, Algebra, Balance, Self, or Obama they all exist as arrangements and connections of neurons in your brain. When a song is stuck in your head that's a physical thing that's happening to a some neurons in your head. But, so is everything else your thinking about.

These networks also connect to and build off of other abstractions so ((((peanut) + butter) + Jelly) + Sandwich) is built from more than one of these networks. Try and think of a pile of peanut butter next to some grape jelly. No problem it's brown next to purple.

Now try and do that for peanut butter next to a peanut butter and jelly sandwich. Most people feel something odd happen on the second one because they are reusing the peanut butter network for both of those. For me, I can't focus on their colors at the same time.

For a great subjective description of this watch: http://www.youtube.com/watch?v=Cj4y0EUlU-Y

Getting back to Consciousness, it's turning on these abstractions by choice. I want to think about Apple so I activate the Apple network and suddenly experience Apple. And it's the same network as if I had just read the word or saw a physical Apple. Language also get's mapped to this which is part of why philosophers think about thinks like platonic ideals yes there is an ideal Chair it exists and it's the Chair classifier in your head.

[ma2rten]

We have different ideas of what consciousness is. Like I said I don't even know for sure if you process consciousness.

http://en.wikipedia.org/wiki/Problem_of_other_minds

I don't think it makes any sense to talk about this any further, we might as well be speaking different languages. Otherwise you could also comment here, http://news.ycombinator.com/item?id=3747462. Naveensundar did a better job explaining it.