mfb said:
How do you simulate a brain? You simulate the behavior of every component - every nucleus and electron if you don't find a better way. You do not need to know about neurons or any other large-scale structures. They naturally occur in your simulation. You just need to know the initial state, and that is possible to get.
I agree with this entirely.
But there is one unexpected piece of neural functionality in the human brain that we will need to replicate: the ability to hold a relatively large amount of information (at least dozens of bits) in a single state. We know that such functionality exists because we are able to be "conscious" of complex concepts, such as the image of a tree, and such objects cannot be summarized in just a few bits - the number that can normally be encoded into a single dynamic state.
But, of course, we do know of devices that can do this - and devices which have the potential to make good (Darwinian) use of information in this form.
On the other hand, I do not see brain functionality that could not be replicated by conventional AND/OR/NAND/NOR gates - even to the point of having the replication report that it is conscious. But what would the purpose of reporting that you are conscious if you are not? Where would the concept of "consciousness" even come from if it didn't exist within social beings? The fact is, we really do have conscious experiences - we aren't just making it up. And, as evidenced by the fact that we can talk about it, that consciousness has the potential to influence our actions.
I will add the argument for how many-bit consciousness compels a many-bit state, though it has fallen on skeptical ears before. Perhaps I can do better this time.
If you are describing something that requires 50 bits, having only 25 of those bits doesn't describe that something. You need all the bits. So you need some way of associating those bits - a way to define which 50 bits stored in this universe are to be the symbolic description of that something. Let's say you use a bunch of logic gates (NAND,NOR,AND,OR) or the presumed neural equivalent. So you have 50 bits of input wired into these gates. But no where in that circuit is all 50 bits, no where in that circuit is the full 50 bits-worth of information associated so that the physics can know there is to be conscious of. For example, you can compute whether the number of 1 bits is odd or even. This will give you a single bit, and therefore a single state, that is dependent of the 50 bits, but obviously, it does not describe the original object.
So how do you associate 50 bits without loosing their value? There is only one physical process for doing this - and being on the Physics Forum should mean that I don't have to say what that is.
Now I am leaving out a piece of this. Associating the bits simply provide one essential element of consciousness, it doesn't "explain consciousness". Fully explaining consciousness has its limits, but if you have followed this so far, there is further to go. Our conscious awareness is very centered around being human, but the basic process required to generate it (superpositioning) is a ubiquitous physical process. It is reasonable to presume that there is a fundamental "consciousness", and that this is implement in the human brain for Darwinian "purpose" with the result being "human consciousness". One more step, made by Penrose, though not in these words: in theory, there is a limited amount on information in the universe - or, in the least, everything that we know about the universe is consistent with there being a finite (though very large) amount of information. Let's create a side universe for ourselves, one with enough flash memory to store a complete description of our universe, and we will make a backup copy of our universe in that flash memory. The question then becomes, how is that backup different from our real universe? That copy will include the full information about humans, but there will be no consciousness. Taken more broadly, all the information about our universe does not make our universe. There is a "reality" element which is the actual physics.
Obviously, not a full explanation. I don't have that.