Has consciousness been fully explained?

[PixyMisa]

As far as I can tell, Piggy is suggesting that the programmers cheated and programmed in a behaviour of saying "I have a Sofia". However that is just altering the thought experiment, nothing is programmed in but the physical interactions of the components of the brain and the sense data.

Yeah, but that's not remotely how a computer simulation of a space mission works. If it were, Florida would be littered with crashed Saturn Vs.

 

[Meed]

The brain doesn't function strictly as a Turing Machine, of course; it's far too messy. It is however a computer. As such, a Turing Machine can approach its function arbitrarily closely.

Arbitrarily closely? What are you talking about?

I have merely pointed out that consciousness displays no magical properties and requires no magical processes. This seems to upset people.

I haven't seen anyone in the thread claim consciousness has "magical properties".

 

[Robin]

Yeah, but that's not remotely how a computer simulation of a space mission works. If it were, Florida would be littered with crashed Saturn Vs.

That is right. A simulation where the expected output was preprogrammed in would be of no interest or value.

 

[Robin]

I haven't seen anyone in the thread claim consciousness has "magical properties".

But could the external behaviour of a human be modelled by a computer program that modelled the interactions of the components of the brain?

 

[Robin]

I haven't seen anyone in the thread claim consciousness has "magical properties".

Or to put it another way - is everybody here claiming that the components of the brain behave exactly as physics says they should?

 

[PixyMisa]

Arbitrarily closely? What are you talking about?

To an arbitrary degree of precision and accuracy; to approximate as closely as is desired, necessary, or physically meaningful; e.g. simulating a physical system to a precision beyond any of the Planck units is a waste of time.

I haven't seen anyone in the thread claim consciousness has "magical properties".

Perhaps not - that's more an Interesting Ian thread, and he's not around any more - but there's a recurring theme that it requires magical processes.

 

[my_wan]

There is no difference. And the statement is accurate. The additional words are merely emphasis. The key word is function. Of course I realise that the Turning machine has to be functioning. That's why I said so. If PM were contending that there were any other necessary element required apart from the function of the Turing machine, then he's had endless opportunities to state as much.

[Bolding mine] I have to take exception to this. The compared statements for which no difference is claimed here, as provided by RD from westprog's post, are:

1) PM's position is that the consciousness of the brain springs from its function as a Turing machine.
2) PM has consistently made the claim that consciousness springs entirely from the function of the brain as a Turing Machine - and nothing else

[Bolding original to RD]
I can accept 1) as a characterization of my own perspective, but not 2). I perceive, but can't know westprog's actual reasoning, that it comes from a misunderstanding of the nature of emergence. I have previously explained how emergent properties are in principle predictable variables, not some intrinsically new property that magically appears in system ensembles.

RD has been demanding a well defined distinction between the mechanistic models used to describe aspects of consciousness and how it differs from consciousness, so I'll provide the principles in terms of emergence and show how it is consistent with the definitions of Myriad and others. This also goes toward the limits of the Church-Turing thesis, and why such a thesis can be valid in all subsystems of a system when not valid in the system itself.

I'll use rogue waves as an example. At the level of individual water molecules you have linear translations through space, mean free path, wrt speed and density. They are simply bouncing off each other. In a standard linear wave, water is not even being transported with the wave, just bouncing back and forth between each other as the wave passes them. Normal waves are an effectively linear emergent phenomena.

Rogue waves are characterized not by a linear compression, but a nonlinear sum of a bunch of chaotic randomly interacting smaller waves. Now, in spite of the individual molecules still being characterized by linear spacial translations, we must model the nonlinearities and treat the emergent wavelets as new entities, and then model the interactions of these new emergent entities to produce a complete new nonlinear emergent entity, created from underlying emergent entities, called a rogue wave.

Now the Church-Turing thesis is limited to ‘explicitly stated rule’, like the collisional rules of the individual water molecules. Yet for the rogue wave we are dealing with a hierarchy of emergent properties, where emergent properties are interacting with emergent properties to produce new emergent properties. It is distinctly nonlinear, and requires input modeling that lacks explicit values as defined by the Church-Turing thesis. Yet foundationally they remain the product of explicit linear mechanistic molecular collisions.

So consider the definitions provided by Myriad and others, where the specific of any given calculation approach in our brains is modified, or evolves, in accordance with our past history and experience with similar such calculations. We may even make tactical changes in how we arrive at a calculation midstream, by noticing patterns during the process of that same calculation. This is a distinctly nonlinear approach to calculations, which don't 'directly' model well, or at all, in a Turing machine. Yet like the rogue wave, the foundational mechanics can remain an ensemble of Turing machines.

Thus there is nothing wrong with describing non-intelligent, non-conscious, linear mechanistic machines as 'operationa'l elements of consciousness. It simply isn't valid to accuse people, that speak of these operational elements as elements of consciousness, of claiming these elements are themselves conscious, period. It's a "Composition fallacy" to accuse people describing composition of a Composition fallacy, when the claimed fallacy was your own composition.

Does this qualify as a well defined distinction RD? Does anybody have any questions not made clear here?

 

[PixyMisa]

I think I agree with you, but I'm a little unclear on what the "nothing else" means.

Turing Machines aren't inherently conscious; it's the algorithm that's running that may be conscious, if it has the right combination of operations. My point is that Turing Machines can be conscious; that everything the brain does can be explained in those terms, and can be emulated or simulated with the same results.

A Turing Machine is a mechanism for running algorithms - any algorithm. If Westprog's "nothing else" means the Turing Machine and no algorithm, then that's just silly. But I'm not sure what else it might mean. Turing Machine and algorithm but no data? That's silly too.

 

[PixyMisa]

There is no difference. And the statement is accurate. The additional words are merely emphasis. The key word is function. Of course I realise that the Turning machine has to be functioning. That's why I said so. If PM were contending that there were any other necessary element required apart from the function of the Turing machine, then he's had endless opportunities to state as much.

If you think that something else is required, please tell us what and why.

You've had endless opportunities to do so.

 

[PixyMisa]

So consider the definitions provided by Myriad and others, where the specific of any given calculation approach in our brains is modified, or evolves, in accordance with our past history and experience with similar such calculations. We may even make tactical changes in how we arrive at a calculation midstream, by noticing patterns during the process of that same calculation. This is a distinctly nonlinear approach to calculations, which don't 'directly' model well, or at all, in a Turing machine. Yet like the rogue wave, the foundational mechanics can remain an ensemble of Turing machines.

I'm not sure what the problem is with running nonlinear systems on a Turing machine. I do it all the time. Sometimes even inadvertently.

 

[Meed]

Although I've only been a graduate student / research assistant for a couple years, this is my field. I'm not trying to be a dick with all my corrections and objections -- I'm trying to clear up misconceptions that tend to be a result of news articles that over-hype or misunderstand scientific findings on occasion. I am wrong sometimes and would like it to be pointed out when I am. If you're right, you're right regardless of my background and occupation. But news stories and TED talks aren't a good source to demonstrate it. Journal articles are a good source.

You misunderstand the principles of what's being monitored. The brain wave game controller uses a low grade MRI, while the dream recorder uses fMRI, where the "f" in fMRI means functional.

Brain wave game controllers measure brain wave frequencies. The MRI game controller (I think you're talking about "Epoc") uses MRI and isn't a brain wave controller, so I assumed that wasn't what you were talking about. I was actually not aware of Epoc until I looked it up. The idea behind it is very cool.

Why is fMRI functionally the same thing as MRI? Because ideally what you want to measure is neither, but instead the locations of activation of individual neurons on a global scale.

Ideally you might want to, but we can't measure the activation of individual neurons, just general brain activity that is assumed based on things like oxygen distribution. Human neurons are (to my knowledge) too small to observe in action.

That's exactly what I noted first when I originally mentioned it, but neither has it been a complete failure.

Well, we don't know do we? Maybe you do, but I'd have to read the details of the study to know to what extent it was a success or failure.

I also explained why some limitations may remain, even if you had exquisite information about every single neural activation throughout the brain. Such an explanation goes to explaining the fine scale randomness of brain structure as discussed here before.

Huh?

 

[my_wan]

I'm not sure what the problem is with running nonlinear systems on a Turing machine. I do it all the time. Sometimes even inadvertently.

The very fact that the nonlinear system described, rogue waves, was a product of Turing machines entails that I made no such claim there was a problem. That's also why I claim there is no problem describing linear mechanistic machines as 'elements' of consciousness when they aren't themselves conscious in any human sense.

I suspect this issue came up when I said:
"This is a distinctly nonlinear approach to calculations, which don't 'directly' model well, or at all, in a Turing machine."
When I put 'directly' was due a recognition that Turing machines can iteratively produce such nonlinear effects, but can't be used to define the initial inputs, per definition of a Turing machine. The "or at all" qualifier is not a claim it can't be done, but a recognition that the Church-Turing thesis pertains only to Turing machines, thus fails to 'prove' that all possible nonlinear systems in nature are reproducible on Turing machines. In spite of extensive examples of nonlinear behavior created with Turing machines, it doesn't include the 'all possible' condition mathematical proofs are predicated on.

Personally I think Turing machines, at the very least as physical ensembles (many acting together), is capable of 'all', but must recognize the difference between my well informed prejudices and what is proven. This has corollaries in foundational physics, where so far we can't pin QM down to a fundamental set of physical Turing machines. I have well informed prejudices there also, but it doesn't make them facts. And I must recognize my well informed prejudices for what they are, and not confuse them with facts.

One 'fact' does remain though, many examples exist in which to fully justify talking of inanimate, non-conscious, Turing machines as elements of consciousness, without assigning those elements human consciousness of their own. Doing so in a limited manner is a perfectly valid conceptual device. The claim that they contain elements of consciousness is not a claim they posses consciousness.

 

[westprog]

What proof?

You haven't offered any proof with regard to your claims regarding the brain.

Like cornsail and westprog, I searched your links (and other sources I found elsewhere) and failed to find that claim.

Not only that - some of the links included a direct and explicit rebuttal of the claim.

 

[PixyMisa]

The very fact that the nonlinear system described, rogue waves, was a product of Turing machines entails that I made no such claim there was a problem. That's also why I claim there is no problem describing linear mechanistic machines as 'elements' of consciousness when they aren't themselves conscious in any human sense.

No problem, just wanted to make sure!

I suspect this issue came up when I said:
"This is a distinctly nonlinear approach to calculations, which don't 'directly' model well, or at all, in a Turing machine."
When I put 'directly' was due a recognition that Turing machines can iteratively produce such nonlinear effects, but can't be used to define the initial inputs, per definition of a Turing machine. The "or at all" qualifier is not a claim it can't be done, but a recognition that the Church-Turing thesis pertains only to Turing machines, thus fails to 'prove' that all possible nonlinear systems in nature are reproducible on Turing machines. In spite of extensive examples of nonlinear behavior created with Turing machines, it doesn't include the 'all possible' condition mathematical proofs are predicated on.

Yes, it's one of those annoying cases where we can neither prove the theorem for all cases nor find a contradictory example.

 

[westprog]

Nice link. Thanks.

It may well have been Pixy who posted it originally. Sorry if that's blown your irony meter.

Of course, Church-Turing is only relevant if the brain is considered as a computational device. If it's considered as a real-time monitoring and control device - which is a role it obviously has - then Church-Turing just doesn't apply.

 

[westprog]

I don't really want to try to breakdown the whole exchange, but for example your idea about the significance of "ability to maintain stability" (paraphrasing) applies to many things, such as rocks. You seemed to take this as the claim that there is no difference between cells and rocks.

He's now demanded that I explain to him what the difference is between a rock and a computer chip. I really don't know what is going on there.

 

[PixyMisa]

What proof?

You haven't offered any proof with regard to your claims regarding the brain.

Like cornsail and westprog, I searched your links (and other sources I found elsewhere) and failed to find that claim.

Read Godel, Escher, Bach.

So it's not up to any of us to disprove your claim. It's up to you to prove it.

You can do that -- if it can be done -- by explaining how Church-Turing proves that it is impossible for the brain to do anything at all a computer can't do (which is pretty extraordinary to begin with, since we don't know how the brain does all that it does).

I've already explained this a number of times.

All known computational models are proven to be equivalent to a Universal Turing Machine. Look it up. All the proofs are online.

So, insofar as the brain is a reliable computer, it is equivalent to a Turing Machine too. (We're also limited by Godel's Incompleteness theorems - to the extent that we are consistent, we are necessarily incomplete.)

Okay, now, there's an obvious problem here: The brain is not a reliable computer. No real-world computer is perfect, but organic brains are notably sloppy.

However, we can model any physical system with a Turing Machine too. We can reproduce everything except quantum randomness directly with a Turing Machine, and we can produce pseudo-random numbers that approach perfect randomness arbitrarily closely. And we can produce a simulation that goes beyond the Planck limit in every dimension, so any remaining differences cannot be meaningful.

So, anything the brain does can necessarily be reproduced by a Turing Machine.

Or, you can find a counter-example - a computational model of greater power than the Universal Turing Machine, or - a related question - find a physical process that cannot be modeled algorithmically.

It all comes down to computability. What I am saying follows on from everything we know about computability; what you are saying (and Westprog, don't know about Cornsail) flatly contradicts it.

If I am right then... Nothing changes.

If I am wrong, and you can show this, then you're up for a Fields Medal or a Nobel Prize. Possibly both, because you've just revolutionised mathematical physics.

Until then, we've only got assertions coupled with links which, as far as I can tell, don't actually support the assertion.

Read Godel, Escher, Bach. It will save you a lot of time and misunderstanding. And it's a wonderful book.

 

[PixyMisa]

It may well have been Pixy who posted it originally. Sorry if that's blown your irony meter.

Of course, Church-Turing is only relevant if the brain is considered as a computational device. If it's considered as a real-time monitoring and control device - which is a role it obviously has - then Church-Turing just doesn't apply.

Yes it does. Sorry to deflate your little counterfactual world bubble there, but yes, of course it does.

 

[PixyMisa]

He's now demanded that I explain to him what the difference is between a rock and a computer chip. I really don't know what is going on there.

No. We've merely asked, if you claim there is no difference, why do you use a computer to connect to the internet rather than a block of limestone.

Well?

 

[PixyMisa]

Not only that - some of the links included a direct and explicit rebuttal of the claim.

Sorry, Westprog, but your failure to understand something doesn't change its truth value in either direction.

And there's a big problem with that Stanford article - the machines that are supposedly more computationally powerful than a Universal Turing Machine all work by magic.

That is, they are all either mathematically, logically, or physically impossible.

So, every model of computation ever dreamed up by mathematics falls into one of three categories:

1. It is a provably strict subset of the capabilities of the Universal Turing Machine.
2. It is provably directly equivalent to capabilities of the Universal Turing Machine.
3. It is provably impossible.

Of the three, which are you arguing that the brain falls into?