"Give a computer something outside of its expected parameters and it just does nothing. Give a human or an animal something outside of its parameters and it will experiment with it."

If you completely simulate a given brain brain, atom up- you will end up with an intelligence.
(Learning seems to require the body)
It is not technically hard to do this at the moment, it simply requires more computer power that can be imagined, and is practically impossible.

Similarly, if you simulate it at a rather higher level - neurons, axons and synapses exactly copied you can create an intelligence.

If you argue 'but it won't be proper intelligence, just look like it' - you can make the exact same argument about the French.

There are various sorts of AI that can be conceived.
A) Slavish copies of brains, emulated at different levels of fidelity.
B) Ones patterned after human or animal brains, but not really designed, more taught.
C) The classical view of AI - you write individual routines to store and recall memories, or to recognise shapes.

C) is vastly harder than A) and B).
We can do A today - in principle - it's simply we lack the computer power and scanning resolution.
B is related to A - but it's like taking a developing brain, and inserting more or faster tissue.
C is a whole nother kettle of fish that I question we can get to.

Understanding in detail how the brain works may never be possible, as the functions are not easily seperable into explicit concepts.

As an aside - truly awesome brain podcast: brainsciencepodcast.com/bsp/neuroplasticity-a-review-of-its-discovery-bsp-10.html
This is intended for the interested amateur, and covers fascinating research into how the brain makes us who we are.

As to how accurate a rendition needs to be to 'think' - it doesn't need to be very perfect.
Neurons change their parameters quite dramatically over all states where we agree there is a person.
For example - the behaviour of neurons in an awake 6 year old, a drunk 18 year old, and a somewhat senile 70 year old differ markedly, yet there is still intelligence there.
(speedevil, Wed 1 May 2013, 17:02, Reply)