What I don't get here is that you keep skimming over the point that a Turing machine + suitable interface doesn't work. Can't work. It won't be a Turing machine any more. It will be a different device altogether.
I know that you keep coming up with ways to avoid this. Yes, a perfect implementation of a Turing machine isn't possible in the real world. So what? The same applies to any design or concept. If we were to use that as an approach, we wouldn't be able to reason about any system.
What makes the concept of the Turing machine useful is that we can make predictions about computations. These predictions are of great practical value. We know that we can launch our Pascal computations into the time-sharing computer, and not worry about implementation details or interaction with the world - and be sure that the program which takes an hour will give exactly the same result as one that take a millisecond.
This is clearly not the case with the replacement neuron. To talk blithely about coupling is to miss the point that a coupled Turing machine is not a Turing machine, and the reasoning we use about Turing machines no longer applies. A Turing machine is, by definition, a closed, non-interacting system. The people who design computers and operating systems have to go to great lengths to provide environments where programs could operate as if they were Turing machines. In almost every case, the computer and operating system which runs the programs has to use a different model, because the Turing model isn't appropriate for running a computer. I gave a link to a paper describing the issues involved in coping with these issues.
So when describing a device which can replace a neuron in a human body, the Turing model is simply irrelevant. Turing-style programs are designed to work as closed systems. The neuron is designed to be open, time-dependent, asynchronous, reactive. The Turing model is of no help in understanding or replacing neuron behaviour.
I just realized your problem. You have a fundamental misunderstanding of the nature of mathematics in general and in particular computer science.
The fact that no turing machine actually exists isn't irrelevant. It illustrates that
any mathematical description is merely that -- a description.
When pixy says that a turing machine can be conscious, he isn't saying that the magical fairy turing machine in abstract land can be conscious. That is absurd. Abstract land doesn't
exist, unless you subscribe to some very loony schools of philosophy. Reality exists.
What he is saying is that
any system which satisfies the behavior constraints we call turing complete -- idealized by the definition of a turing machine -- we can be conscious. That is a very, very, very different claim than what you seem to think he is saying.
This is equivalent to saying "anything that is round -- idealized by the definition of a circle -- can roll." Period. No different. "Round" and "turing equivalent" are the same kind of thing in human language -- a description, a
mathematical description, of a
class of behaviors of
real stuff, regardless of what one thinks of the ultimate "nature" of that stuff.
"Round" is the class of behaviors that lead to a system interacting with other systems in a way that is mathematically isomorphic -- after much filtering is done by the observer -- to a circle. There is a location within the system that each location on the surface of the system is equidistant from. The idealized description is what we call a circle -- so what. Nothing perfectly round exists. No circle actually exists. Hmmm... kind of like no perfect turing machines exist? Yet you don't complain when people say "anything circular can roll," do you?
Likewise, "turing equivalent" is a class of behaviors that are mathematically isomorphic, after much filtering done by the observer, to a turing machine. <insert formal definition of turing machine>
The thing that allows a round object to roll is how near it is to the idealized circle. But you still need a hill, gravity, etc, or else nothing can roll. But nobody bothers to argue this point because it is utterly stupid, given that when someone says "a circle can roll" everyone knows what they mean -- we have grown up and this concept is so familiar it isn't an issue.
To pixy and I, who are very familiar with computer science, saying "a turing machine can do X" is not controversial in the least. We
understand that the implicit meaning is "a turing equivalent system can do X, probably with the help of lots of stuff that isn't necessarily part of the ideal turing machine, just like a round object can't roll without the help of many things that have nothing to do with the ideal circle."
Why is that so hard for you to understand? Turing machines don't really exist. Circles don't really exist. When people speak of either they are referencing reality, not fairy land.
