On Consciousness

[steenkh]

Is this statement true? Some years ago a computer-code driven "neural net" had surrounding code which contained various variables. Through multiple training runs the code in operation set those variables to the best values, that is, trained the neural net.

First of all, it is possible to built the neural network in hardware, so no program is needed. Secondly, you could regard the neural network program as the operating system.

What is important here is that the neural network is not programmed to do what it does: there is no sequence of instructions that tells it how to achieve its purpose.

You could regard the system of "rewards" and "punishments" that is used to program the neural network as a "program", but then humans and animals are also run by programs in the same way.

 

[Belz...]

My point is that a bunch of 1's and 0's doesn't fly. A collection of subatomic particles does.

I would invite you to try to think of particles as 1s, for example.

If a simulated plane can't fly, then why would a simulated brain think?

This has already been explained to you, so your confusion is not understandable. Thought is computation, computer simulation is computation. Therefore simulated thought is thought. Do you have a problem with this ?

Do you think it's possible we're in a simulation?

I think it's terribly unlikely but not entirely impossible.

I'm questioning the assertion that brains are computers, that machines are conscious, that simulated brains will be conscious. It all rests on the belief that consciousness is computational, which is simply asserted as if it's been proven.

Ok, then. You say:

Solitaire and chess don't refer to physical things. They're abstract sets of rules and goals.

But they do refer to physical things: namely actions taken with objects or representations of objects themselves in physical form or media. Thought also does not refer to physical things except in this sense. What's the difference ?

What is thought, really, if not a set of calculations ? I think there's plenty of evidence that thought/consciousness is computational, but I'm willing to hear your counter.

Please don't cut off my quotes. You've done that before in this thread.

Did my cutting a quote somehow change the meaning ? If so, then I'll try to be careful in the future. Otherwise, no.

 

[dlorde]

When you look at the set that makes up "computer", they are all essentially collections of mechanical binary switches. That's where the computing occurs, would you agree?

No. Analog computers use continually varying quantities to calculate, not discrete values. Where are the binary switches in a hydraulic computer that calculates using relative volumetric displacements? or the cams, wheels, cogs, and levers of the gun control computers that calculate using relative motion?

Do you think a brain is a collection of binary switches?

Forget binary, it is incidental. Binary is just a convenience for electronic computers, that operate by switching a set voltage.

I'm not aware of any examples of computers that have been given that aren't binary collections of mechanical switches.

Then you're not paying attention. See above.

So far, what seems to convince people that a brain is a computer is that it computes. That just establishes it's like a computer, does one thing that a computer also does.

I've said all I want to say about this, it's basically a semantic issue.

You left off the mercury thermostat I asked you about. It controls a complex heating/cooling system. Are you comfortable claiming such a thermostat feels? If such a simple device can feel (or think), than just about anything can.

Yes; I'm comfortable with it. It senses changes in the environment and changes its behaviour accordingly. I wouldn't say it thinks.

As to your question, I would say that something can think if it has a sufficiently advanced nervous system. I know that's a bit of a dodge.

Don't dodge, think. What's the difference between thinking behaviour and unthinking behaviour?

What do you mean by "respond"? A rock can "respond" to a change in environment by warming or cooling. Iron atoms "respond" if you bring a magnetic field into the environment. This is what I mean by a trivial definition that results in almost everything feeling/thinking.

Yes, it's a reasonable point; water can be said to feel the influence of gravity and respond by flowing downhill. It's a question of where we make our distinctions and why. I make mine where stimulation or activation of a sensor is associated with a behaviour modification. In a thermostat, activation of a temperature sensor results in a switch from one behavioural state to another. Clearly, it is a trivially simple example of feeling.

The subjective feelings we experience. Pain feeling bad is a good example.

Feelings are feelings? how useful a definition is that?

By defining them as subjective experiences that we have, you seem to suggest only we can feel - did you mean that?

Well, we have our own conscious experience to guide us, but it's hard to formulate that into a definition.

No-one said it would be easy. It's hard because you have to think and make decisions about what you mean, instead of vague hand-waving.

That doesn't mean I cant critique other definitions. If someone claimed consciousness was just adding numbers together, I could definitely say they were wrong.

But you couldn't explain why. It's not critique without explanation, it's just unsupported assertion.

An artificial brain would be like a brain, but made out of non-organic components. I don't see why it wouldn't be conscious.

Like a brain, but non-organic, and, as you said, you'd accept transistors, binary logic, and a number of separate boxes... so in what ways 'like' a brain? would it compute?

When it can be proved that computers can think/feel/and are conscious, I would be sympathetic to such a claim.

Can you prove that you are conscious? if so, how?

 

[dlorde]

What's the difference between simulated thinking and consciousness and real thinking and consciousness?

Obvious: one is simulated and the other is real.

OK, I think that speaks for itself; time to move on.

 

[Jeff Corey]

...I'm assuming nothing except that brains are conscious. I'm questioning the assertion that brains are computers, that machines are conscious, that simulated brains will be conscious. It all rests on the belief that consciousness is computational, which is simply asserted as if it's been proven...

I agree, with the qualification that some brains are conscious, some of the time. .

 

[PixyMisa]

Brains are computers. We can see that when we open them up and poke them; there they are, computing away.

That's not in question.

The question rather is, are brains more than computers? And as far as we can tell, that's simply not physically or mathematically possible.

I agree, with the qualification that some brains are conscious, some of the time. .

Yes to that. Consciousness is something the brain does, not something it is.

 

[Fudbucker]

OK, I think that speaks for itself; time to move on.

I'm not being glib. I'll make a simple logical argument:

1. A physical process is not the same as it's corresponding simulation.
(simulated digestion is not the same as digestion, simulated photosynthesis is not the same as photosynthesis, simulated flying is not flying, etc.)
2. Consciousness is a physical process.
3. Therefore, consciousness is not the same as simulated consciousness.

Now, what part do you disagree with? Do you think consciousness is NOT a physical process?

 

[PixyMisa]

I'm not being glib. I'll make a simple logical argument:

1. A physical process is not the same as it's corresponding simulation.
(simulated digestion is not the same as digestion, simulated photosynthesis is not the same as photosynthesis, simulated flying is not flying, etc.)

Those are two different statements, and the second statement does not follow from the first.

 

[Belz...]

I'm not being glib. I'll make a simple logical argument:

1. A physical process is not the same as it's corresponding simulation.
(simulated digestion is not the same as digestion, simulated photosynthesis is not the same as photosynthesis, simulated flying is not flying, etc.)
2. Consciousness is a physical process.
3. Therefore, consciousness is not the same as simulated consciousness.

Now, what part do you disagree with? Do you think consciousness is NOT a physical process?

1) is wrong.

 

[steenkh]

1. A physical process is not the same as it's corresponding simulation.
(simulated digestion is not the same as digestion, simulated photosynthesis is not the same as photosynthesis, simulated flying is not flying, etc.)

This is where you go wrong. A simulation can can result in the same as the real thing. I have once been working on a tape recorder emulation, and it actually could record and play back, even though it was clearly not a physical tape recorder.

 

[dlorde]

This is where you go wrong. A simulation can can result in the same as the real thing. I have once been working on a tape recorder emulation, and it actually could record and play back, even though it was clearly not a physical tape recorder.

Yup, BTDTGTTS.

Coming next: circular semantic games, e.g. 'if it gives the same results it's not a simulation', etc. Rinse & repeat.

 

[Mr. Scott]

Ah the "magic mathematics bean" argument.

Real magic [snip] refers to the magic that is not real, while the magic that is real, that can actually be done, is not real magic. (Lee Siegel via Dennett)

When we talk about, say, a magic bean of cosmic consciousness, we are talking about a kind of "real magic" that is easily proven to not be real. Mathematics magic, on the other hand, which is not real magic, is nevertheless real enough to actually work (which gave you computers, the Internet, cell phones, reaching the moon, wing suits, neural network simulations).

 

[Fudbucker]

No. Analog computers use continually varying quantities to calculate, not discrete values. Where are the binary switches in a hydraulic computer that calculates using relative volumetric displacements? or the cams, wheels, cogs, and levers of the gun control computers that calculate using relative motion?

You have a good point.

Yes; I'm comfortable with it. It senses changes in the environment and changes its behaviour accordingly. I wouldn't say it thinks.

Then my worries about "feel" becoming trivialized are true. If such a simple mechanical device can "feel" then pretty much anything can. Either nearly everything "feels" or your definition is wrong. I think the latter.

Don't dodge, think. What's the difference between thinking behaviour and unthinking behaviour?

Thinking, obviously.

Yes, it's a reasonable point; water can be said to feel the influence of gravity and respond by flowing downhill. It's a question of where we make our distinctions and why. I make mine where stimulation or activation of a sensor is associated with a behaviour modification. In a thermostat, activation of a temperature sensor results in a switch from one behavioural state to another. Clearly, it is a trivially simple example of feeling.

Or not an example at all. The definition of feeling must articulate what it's like to subjectively experience something, such as joy. If a thermostat can feel, you should be able to ask Nagel's question: what is it like to be a thermostat? Such a question, with regards to thermostats, is ludicrous.

Feelings are feelings? how useful a definition is that?

Feelings are subjective experiences. It's not enough to say that hunger informs an organism that it needs to eat. Hunger also has a feeling that goes along with it: the experience of being hungry.

By defining them as subjective experiences that we have, you seem to suggest only we can feel - did you mean that?

No, but organisms with nervous systems are the only things we know that are capable of feeling anything.

No-one said it would be easy. It's hard because you have to think and make decisions about what you mean, instead of vague hand-waving.

It's not hand-waving. We don't know what the cut-off is for the ability to think. This doesn't mean we can't declare that such and such can think or not. What is the exact cut-off for "rich"? An income of $200,000? $500,000? Impossible to exactly say. However, that does not stop us from confidently declaring that Warren Buffet is rich while a migrant farmworker is not.

Like a brain, but non-organic, and, as you said, you'd accept transistors, binary logic, and a number of separate boxes... so in what ways 'like' a brain? would it compute?

If it was an artificial brain, it would compute like an organic brain. It would do everything an organic brain does.

Can you prove that you are conscious? if so, how?

To myself? I can't be wrong about it, else the word has no meaning. Prove it to you? My nervous system matches the kind of nervous systems we know can produce consciousness. My behavior corresponds to behavior seen in conscious beings.

Would a computer be conscious if it passed a Turing test? Is Data conscious? I don't know. I would err on the side of caution though, and treat it as if it were.

 

[Fudbucker]

Yup, BTDTGTTS.

Coming next: circular semantic games, e.g. 'if it gives the same results it's not a simulation', etc. Rinse & repeat.

Really now. When scientists simulate the early conditions of the Big Bang, do you believe there's a tiny universe in the computer? Do you think they go in a bomb shelter when they model nuclear explosions? Physical processes are obviously not the same thing as simulations. I don't know what to say to those who claim they are. There's such a disconnect there, it's unbridgeable. Maybe this advice: don't cry over the people you murder in Grand Theft Auto. It's all right.

 

[AlBell]

If it is an artificial neural network running on a computer, the program runs the neural network, so technically it is still a program.

So I am correct in my understanding. Thanks.

If it is a physical neural network, then no, there is no "program."

I'll disagree if you are referring to brains which are programmed first by genetics, then by sensory input and private behavior, that is, mulling over what one understands and how it all fits together. Those activities provide the various weightings & connectivity.


Would you or some other comp.lit take a shot at answering my previous comment/question?

"Regarding cosciousness, I agree that human level public behavior could be demonstrated by a computer controlled system of sensors and moving servers.

I, and others I suspect, quibble with private behavior, since that *is* my consciousness as I understand it. How does your computer system manage that, and how would anyone ever know it did?

As an aside my conscious thoughts are streams of words, which of course at times are interrupted by external stimuli needing attention."

 

[Mr. Scott]

The same reason real plants can perform photosynthesis and simulated plants can't. A physical brain produces consciousness.

OK, answer this one:

Let's consider emulation. We know we can make perfect emulations of, say, an Atari, on a PC. With no physical Atari, it plays PONG precisely the same as a physical Atari. We can say Pong or substrate indifferent.

Say, in the future, we are able to read the complete circuitry of neurons in the human brain (each of which can be emulated now on computers BTW) and create a computer of transistor switches and microcode that emulates the human brain (just as a PC program emulates an Atari), then attach it to a robot capable of seeing, hearing, and moving much like us.

Wouldn't it think and act identically to a human, just like a PC running an Atari emulator would run identically to an Atari? Wouldn't it be conscious?

What could it be about the brain substrate that could produce consciousness, when an identical network of switchs on computer chips, could not? The claim of substrate dependency would seem to be extraordinary, so where is your extraordinary evidence?

 

[!Kaggen]

When we talk about, say, a magic bean of cosmic consciousness, we are talking about a kind of "real magic" that is easily proven to not be real. Mathematics magic, on the other hand, which is not real magic, is nevertheless real enough to actually work (which gave you computers, the Internet, cell phones, reaching the moon, wing suits, neural network simulations).

Mathematics "works" no better than any other language. Language does not make things "work", it help us abstract a physical process or thing so that it can be communicated and then repeated or described by others or ourselves at a later date. The word "chair" for instance allows us to predict that when someone advertisers a chair for sale it will behave like a chair when we purchase it.

There is certainly a "magic" in language in that the more effective it is in the above task the more difficult it becomes in differentiating between the abstraction and the thing abstracted.
This can lead to the religious mystical idea of human language being related to a cosmic creative process.

Think:

In the beginning was the word
Logos
Plato
Pythagoras

Comp. lit. is simply a continuation of this tradition.

 

[trebor]

My nervous system matches the kind of nervous systems we know can produce consciousness. My behavior corresponds to behavior seen in conscious beings.

Can you prove your brain is not just simulating consciousness?

 

[rocketdodger]

How does your computer system manage that, and how would anyone ever know it did?

The theory I am proceeding on is that it isn't feasible for a machine to display human level public behavior without also relying on "human-ish" private behavior to do so.

There simply is no reason to think that a machine would ever be talking about how it experiences red unless it actually did experience red.

People like to latch on to the idea that you could just "program" a machine to pass a genuine Turing test, but that isn't realistic. That will never happen.

 

[AlBell]

The theory I am proceeding on is that it isn't feasible for a machine to display human level public behavior without also relying on "human-ish" private behavior to do so.

So you are relying on a successfully passed Turing test as complete proof your computer is human level conscious. As I suspected. Thanks.

There simply is no reason to think that a machine would ever be talking about how it experiences red unless it actually did experience red.

Agreed that coding could differentiate the correct wavelengths we 'experience' as red. And none of us can ever verify how others 'experience' red either.

People like to latch on to the idea that you could just "program" a machine to pass a genuine Turing test, but that isn't realistic. That will never happen.

Our pixy-bot sometimes gets close.