The Hard Problem of Gravity

[PixyMisa]

Pixie,

I'm trying to get some clariity with these terns: "conscious" and "aware".

As far as I know, my body and brain are engaged in hundreds of self referential loops of which I have no awareness. (No perception of them and I'm totally oblivious to these processes.)

You can't access the internal state of these loops, but your mind is entirely dependent on their functioning, yes.

Then there's this symbol manipulation I call "self-awareness" which is also a lot of loopiness.

Yep, that's inherently loopy.

I seem to be aware of this awareness, though I suspect I'm not aware of the of what's actually going on but just a surface process or even a derivative.

Right. It's a self-referential loop - we can see that it has to be by the results it can produce - but exactly what it is doing is more complex. For example, we know from experiment that it's not causally efficaceous, at least not directly. When you consciously decide to take an action, your awareness of that decision follows after the motor nerves firing to begin the action itself.

Now the dumb question for clarity's sake:

Not dumb at all!

Is SHRDLU aware?

SHRDLU can answer questions regarding its environment. Can you do that without being aware?

Is SHRDLU self-aware?

SHRDLU can answer questions about its actions and reasoning. I'll state categorically that you can't do that without being self-aware, without having those self-referential loops.

Is SHRDLU aware of its awareness?

I think that's largely the same question as the previous one, but with a slight twist. SHRDLU doesn't have the vocabulary to describe its self-awareness, but does clearly apply it to answer other questions. ("Because you asked me to.")

Does SHRDLU have a subjective experience of selfhood?

Again, I'll flip this around, and say if it can answer questions regarding its reasoning and actions, how can it not have a subjective experience of selfhood? What else is this subjective experience supposed to be?

BTW I'd reccomend folks read Hofstadter's I'm a Strange Loop.
GEB is fine but somewhat dated. LOOP makes a deeper investigation into what we call selfhood.

I haven't actually read I Am a Strange Loop, so you're ahead of me there!

 

[Apathia]

Thanks Pixy.
So by your reckoning there are language and counseling bots that exhibit all there is to self-awareness.

Do we regard them as selves then, in the way we regard ourselves as selves?
What of a high level AI NPC in a game.
Have I killed someone when I blast him away?

I would have to say that if SHRDLU has the same "inner" experience of selfhood I do. And is aware of herself as a self, then I have must have more empathy toward her than a do a misquito.

But I know I'm dealing with an element of subjectivity here.
And it seems that that's a very problematic and perhaps unwelcome thing to bring to the discussion.
The only content that can be analyzed are loops of self referencing circuitry.
What the heck could anyone mean by being a self otherwise?

RIP SHRDLU. They pulled the plug on you!

 

[PixyMisa]

Thanks Pixy.
So by your reckoning there are language and counseling bots that exhibit all there is to self-awareness.

It's a complex question. Certainly they do not exhibit the full spectrum and complexity of human self-awareness. SHRDLU for example lacks the vocabulary to consider the question directly.

Do we regard them as selves then, in the way we regard ourselves as selves?
What of a high level AI NPC in a game.
Have I killed someone when I blast him away?

I'd say that it's not consciousness itself that we care about, but the complexity of that consciousness. So the subroutine that manages the actions of an NPC might be conscious (in the sense we are using here), but less so than, say, an ant, and also possessing some other properties that make us less inclined to worry about snuffing it out. For example, it can be indefinitely reproduced, and also saved and recreated identically. Would we worry so much about killing even people if we knew for certain that there was reincarnation?

I would have to say that if SHRDLU has the same "inner" experience of selfhood I do. And is aware of herself as a self, then I have must have more empathy toward her than a do a misquito.

Well, SHRDLU has some sort of inner experience, since it can talk about it. But a mosquito would also have an inner experience - it just lacks the vocabulary.

But I know I'm dealing with an element of subjectivity here.

Not exactly. We're talking about subjectivity, but the existence of subjectivity is a well-established objective fact.

And it seems that that's a very problematic and perhaps unwelcome thing to bring to the discussion.
The only content that can be analyzed are loops of self referencing circuitry.

But that's just what I (and of course Hofstadter) are saying that subjectivity is.

What the heck could anyone mean by being a self otherwise?

Exactly.

RIP SHRDLU. They pulled the plug on you!

If you can find a copy of Micro Planner, you can have a SHRDLU of your very own.

 

[yy2bggggs]

And unicorns and fairies are not completely independent of physical reality. On the contrary, they are completely dependent on physical reality.

This sounds to me like a reification. Whereas unicorns and fairies are not real, they are not dependent on anything. Since there are no unicorns, and since there are no fairies, there's nothing there to be dependent on reality. Now, there's a concept of a fairy, and a concept of a unicorn, and those concepts are based on reality--but then again, concepts that extant entities like ourselves have necessarily have to be based on reality--because we're extant entities, and the act of conceiving is a real thing.

Now, in this respect, most definitely--mathematics is dependent on reality, just like fairies are. But what do you gain from claiming that mathematics depends on reality in the same way that fairies depend on reality?

Likewise for mathematics.

Mathematics is the names and symbols. Mathematics is the description.

Mathematics is the logical extension of relationships. The description is a game we play--a means by which we pull extremely complicated relationships that would otherwise be beyond our grasp into arm's length so we can touch it. Since this game is such a useful thing, and since it is almost second nature when we practice mathematics for us to use the symbols and the things we've pulled into our arm's reach, and since that's pretty much all we do when we play the mathematics game, we can pragmatically call the game itself mathematics.

But when push comes to shove, what we're talking about isn't the language--it's not the set of rituals we go through to get the answer--it's the relationships.

It is. Westprog believes that mathematics exists independent of physical reality and that physical reality just happens to perfectly obey the laws of mathematics due to an unknown magic.

That sounds pretty accurate to me. Reality is mathematical because it has relationships--and not just relationships, but relationships that extend in complicated ways, such that we can play our game to pull certain things within our grasp. In other words, physical reality is mathematical because it is symmetrical.

There's no a priori reason why physical reality must be symmetrical. We know physical reality has many symmetries, but we know this through a posteriori means. It's quite a beautiful thing, but I for one can't explain why it should be that way.

In this particular respect, I fail to see what westprog has said that was wrong.

 

[cyborg]

Reality is mathematical because it has relationships

Sigh.

Relationships are synthetic.

There are no "relationships" out there floating about. There are descriptions of groupings of objects that are imposed upon those objects with no obligation for those objects to behave in that manner.

In other words, physical reality is mathematical because it is symmetrical

I don't know this definition of symmetrical you're using.

I for one can't explain why it should be that way.

There are only a limited set of basic possibilities.

Either the behaviour of objects is consistent in which case their is an algorithm of finite length that can describe their behaviour or the behaviour of objects in inconsistent in which case there is no algorithm of finite length that can describe their behaviour.

Law of the excluded middle.

These things are only surprising if one insists on viewing one portion of the infinite spectrum of possibilities are declaring it amazing that it exists.

 

[yy2bggggs]

Sigh.

Relationships are synthetic.

There are no "relationships" out there floating about.

I don't have any clue what you're on about here. Not only am I not claiming that relationships are out there floating about--I have absolutely no idea what that even means.

I don't know what to make of your claim that relationships are synthetic either. If you're talking about the relationship between objects, they are most definitely not synthetic. I don't get to choose how reality interrelates--at best I get to rearrange objects such that certain fun things happen. I may be a master of rearranging, but I'm a slave to physical law--mine is but to learn what the relationships are and how to exploit them.

I don't know this definition of symmetrical you're using.

You're calling it consistency.

There are only a limited set of basic possibilities.

Sure, but one thing you should always be very wary of is your ability to enumerate open ended things, even "obvious" things.

Either the behaviour of objects is consistent

...okay, but consistent in what way? If the same types of things that occur now always occur, that means that things tend to work the same way no matter when you do them. That is a symmetry--in particular, it's a translational symmetry in time. If the way things work here tend to be how they work everywhere else, that is a symmetry in space.

Consistency implies symmetry--there has to be something you can hypothetically do to the system after which everything is still the same.

in which case their is an algorithm of finite length that can describe their behaviour or the behaviour of objects in inconsistent in which case there is no algorithm of finite length that can describe their behaviour.

Alright, not sure what you mean exactly here, so I'll ask a question before deciding whether or not to agree. Would you consider pi's digits to be consistent or inconsistent? And would you consider a description of how to compute the digits of pi, that was in itself finite, to be an algorithm of finite length?

Law of the excluded middle.

What about partial descriptions? How about, say, QM?

These things are only surprising

I don't need it to be surprising--simply not certain a priori would do the trick. I'm not surprised by symmetry. I just don't see how you can rightfully conclude the universe is symmetrical without taking a peek. You seem to be saying the same thing here:

There are descriptions of groupings of objects that are imposed upon those objects with no obligation for those objects to behave in that manner.

So what exactly are you disagreeing about?

 

[cyborg]

I don't have any clue what you're on about here. Not only am I not claiming that relationships are out there floating about--I have absolutely no idea what that even means.

Fine.

I don't know what to make of your claim that relationships are synthetic either. If you're talking about the relationship between objects, they are most definitely not synthetic.

Yes, they most definitely are.

I don't get to choose how reality interrelates

No, but you are choosing how to describe it. You cannot aviod that. Hence synthetic.

Consistency implies symmetry--there has to be something you can hypothetically do to the system after which everything is still the same.

I'm not entirely sure I see how this applies - consistency implies that the same system would behave in the same way every time.

Would you consider pi's digits to be consistent or inconsistent?

They are consistent with the digits of pi. (Not an attempt to make a stupid point).

And would you consider a description of how to compute the digits of pi, that was in itself finite, to be an algorithm of finite length?

Since "a description of how to compute the digits of pi" is an algorithm this has to be the case.

What about partial descriptions? How about, say, QM?

They don't allow for anything new - they exist on the edges of these descriptions which makes them quite intractable.

So what exactly are you disagreeing about?

That reality is mathematical.

That mathematics contains an isomorphism to the behaviour of reality is hardly surprising since physics was constructed for that purpose.

 

[westprog]

You just don't get it.

Mathematics is the names and symbols. Mathematics is the description.

Math is a language. Just like any other language, it is a description.

People don't say "English is entirely independent of the names and symbols we use to describe it." English doesn't float around in the void independently of human experience.

Neither does mathematics.

No, it isn't. A Chinese mathematician using an entirely different set of symbols will do research entirely independently of his American counterpart, and come up with exactly the same results. That's because he's researching into things that are objectively true.

 

[cyborg]

A Chinese mathematician using an entirely different set of symbols will do research entirely independently of his American counterpart, and come up with exactly the same results. That's because he's researching into things that are objectively true.

Er no, the tacit assumption here would be that they're doing mathematics based on the Peano axioms. The jots and tittles of the symbols are irrelevant because they don't affect the logic of the things they represent.

Now tell me:

The Chinese mathematican tells you that the two parallel lines will intersect and the American mathematican tells you that they won't. Who is right?

 

[westprog]

That mathematics contains an isomorphism to the behaviour of reality is hardly surprising since physics was constructed for that purpose.

Mathematics was discovered via the properties of reality. We know that 1+1=2 by counting apples. What was then discovered (and quite early on, with Euclid) was that the truths of mathematics don't need the apples. The apples need the maths.

 

[Darat]

Mathematics was discovered via the properties of reality. We know that 1+1=2 by counting apples. What was then discovered (and quite early on, with Euclid) was that the truths of mathematics don't need the apples. The apples need the maths.

It is the other way around - the apples do not need the maths - they exist whether anyone has any "maths" that models/describes/defines some aspect of their existence. (This holds for any monism based on the premise "something exists".)

Think about it using another way of defining reality: do the apples need the English language to exist?

 

[Darat]

Mathematics was discovered via the properties of reality.
...snip...

And no mathematics was invented as a means to model/describe/define certain aspects of reality.

 

[Nick227]

SHRDLU can answer questions regarding its environment. Can you do that without being aware?

SHRDLU can answer questions about its actions and reasoning. I'll state categorically that you can't do that without being self-aware, without having those self-referential loops.

It strikes me that actually the former is more of a test of awareness. Accessing memories of events seems to me less reliable. Being asked "Do you see the table" to me is inherently better here than being asked "Did you see the table yesterday?"

For me also, if SHRDLU has its full processing capacity engaged with the table then it cannot be aware of the table. There needs to be a sense of perspective, in humans usually generated by concurrent awareness of self or something else in the background. In Strong AI I imagine awareness is really just working memory.

In addition, for me I would need to know that SHRDLU possessed the programming to perform the feats it's being questioned about, or the programming to answer such questions correctly. Otherwise it could simply be answering in a pat way.

I think that's largely the same question as the previous one, but with a slight twist. SHRDLU doesn't have the vocabulary to describe its self-awareness, but does clearly apply it to answer other questions. ("Because you asked me to.")

For me this is getting more complex now, because "aware" can have subtly different meanings in English. Do you mean "Do you know you're aware?" or "Are you aware of being aware?" The latter SHRDLU would certainly struggle with imo.

Again, I'll flip this around, and say if it can answer questions regarding its reasoning and actions, how can it not have a subjective experience of selfhood? What else is this subjective experience supposed to be?

As I read Strong AI, SHRDLU will only have a subjective experience of selfhood if it has the capacity to construct selfhood. If SHRDLU has not the programming to teach itself what is SHRDLU, from the range of objects it is coming into contact with, then it will not have a "visual" sense of self. There are other aspects too.

Nick

 

[PixyMisa]

It strikes me that actually the former is more of a test of awareness. Accessing memories of events seems to me less reliable. Being asked "Do you see the table" to me is inherently better here than being asked "Did you see the table yesterday?"

Why?

For me also, if SHRDLU has its full processing capacity engaged with the table then it cannot be aware of the table.

Again, I have to ask you if you think that actually means something.

There needs to be a sense of perspective, in humans usually generated by concurrent awareness of self or something else in the background.

Why?

In Strong AI I imagine awareness is really just working memory.

I imagine you are completely wrong.

In addition, for me I would need to know that SHRDLU possessed the programming to perform the feats it's being questioned about, or the programming to answer such questions correctly. Otherwise it could simply be answering in a pat way.

I see.

So if Descartes explains to you how he arrived at his cogito, and you realise that all that's really happening is neurons firing in his brain, then he doesn't understand any of this, he's just answering in a pat way?

For me this is getting more complex now, because "aware" can have subtly different meanings in English. Do you mean "Do you know you're aware?" or "Are you aware of being aware?" The latter SHRDLU would certainly struggle with imo.

Yes. It's a language issue - on two levels.

As I read Strong AI, SHRDLU will only have a subjective experience of selfhood if it has the capacity to construct selfhood. If SHRDLU has not the programming to teach itself what is SHRDLU, from the range of objects it is coming into contact with, then it will not have a "visual" sense of self. There are other aspects too.

Read Hofstadter. He actually uses SHRDLU as an example.

 

[Dancing David]

I'm pointing out that objectivity is just a brain state. In a sense, it has no external validity. You can be entirely sensorily conscious and yet have no objectivity. Thus, in considering the validity of objectivity in uncovering the materialist basis of consciousness, I would say...if it gets the job done, great. If not drop it.

Nick

Objectivity follows the same rules under the immaterialist model as well, it is not dependant upon materialsim.

You just don't like it, the sense of self begins with sensation and perception, not the narrative language. Body, sensations, emotions, thoughts and habits.

 

[Dancing David]

That's a very strange thing to say. How do you define objectivity here?

Please don't, it is what Nick227 thinks his existential mumbo flummery can help with.

 

[yy2bggggs]

No, but you are choosing how to describe it. You cannot aviod that. Hence synthetic.

But I don't chose how it's related, and you can't avoid that. An accurate description of reality is one that correctly describes the relationships. The ultimate thing I'm after when I describe reality is making sure I have the proper relationships.

But even this is beside the point with respect to mathematics. No matter what reality is like, in Euclidean geometry, the square of the hypotenuse is exactly the sum of the squares of the other sides. Euclidean geometry defines a particular class of relationships in itself. Even if science and physical reality turns its back on Euclidean geometry, mathematics still cares. It leaves this abstract world untouched, unharmed, because the relationships in themselves are what math is about.

I'm not entirely sure I see how this applies - consistency implies that the same system would behave in the same way every time.

That's translational symmetry in time. That's one of many symmetries in the universe.

They are consistent with the digits of pi.

Okay, but isn't that a tautology?

I'm even more confused about how pi's digits being consistent with pi's digits relates to this:

Either the behaviour of objects is consistent in which case their is an algorithm of finite length that can describe their behaviour or the behaviour of objects in inconsistent in which case there is no algorithm of finite length that can describe their behaviour.

(Furthermore, there has to be a consistency to begin with in order for us to speak of things such as "objects" in the universe in the first place--you don't just get to have objects--there has to be some form of thing with a separable and meaningfully consistent identity first).

They don't allow for anything new - they exist on the edges of these descriptions which makes them quite intractable.

Even more confused. How would QM fit into the "law of excluded middle"? I'm trying to figure out if by "law of the excluded middle" you really mean something I would call "false dichotomy". It really severely looks like reality can be both describable and not, at the same time--that there are things that can be described, and things that cannot be.

I'm generally flexible with the way people describe things. But I'm pretty firmly against the notion of a priori reasoning about the nature of ontological entities.

You said reality was either one way or the other. I claim it not only can be both, but probably is. If you didn't mean to imply an exclusive or to this case, I'm perfectly fine with it. Otherwise, I object.

That reality is mathematical.

Be more specific. Is it your position that reality is not mathematical? That, say, there are no relationships in reality?

That mathematics contains an isomorphism to the behaviour of reality is hardly surprising since physics was constructed for that purpose.

But you do isomorphisms all of the time to get the answers anyway. When you multiply 15 by 15, you may do either of the following:

• 25+50+50+100=125
• 100+25+2*10*5 (a^2+b^2+2ab)
• Make a grid of 15x15 dots, and count them

Either way, the goal is to get 225. But 225 is shorthand, in itself, for 2*100+2*20+5. It's the thing we want because it is a standard form that allows many other kinds of standard games to be played (such as counting using the decimal system).

But what all of this is about is how many dots there are when you arrange them in a square, or how much a pack of 15 flubnars costs at 15 pence a piece, or the momentum of a 15kg ball traveling at 15 m/s. 225 is the thing we want, because that's our goal. 225 is related to the number of dots, or the cost of flubnars, or the momentum of the ball in kg m/s.

Whether you call the games you play to get 225, or the particular relationships between the entities 15, 15, and 225 using this multiplication type relatedness, mathematics, is in itself nothing but semantics. But the real piece of interest is that 15 and 15 under this multiplicative relatedness is related to 225. Or if your problem happens to be different--you may take interest in the fact that the relatedness of interest between 15, 15, and your answer is the multiplicative kind of relationship.

This entire subject of concern is mathematics^dict--the games we play to get there, the fact that we can get there multiple ways, other ways to get there, what game we play to get there--it's all mathematics.

ETA: Left out "where we want to be".

 

[Dancing David]

Thanks Pixy.
So by your reckoning there are language and counseling bots that exhibit all there is to self-awareness.

I think it is a little more complex. first off the system is self referencing, but it is also plastic, it responds and changes.

All of it is acquired through the fuzzy logic of association. Rational thought (verbal cognition) is not the only path. A lot of it comes from pattern matching, language acquistion involves a lot of guessing and intuition.

Which is why idiomatic contruction of language is so fraught with misinterpretation. It involves the consistent application and use of the idiom to understand it's application. So 'bot' is not exactly what it is.

I rememebr three years ago when in middle school, they first used 'cronk' around me, "It is getting cronk up in here." the meaning in that context is "fun" or "party". While the more general meaning is 'high", which reminds me of my youth, there was 'wasted', 'toasted', 'blasted'. 'poop-faced', 'lit', 'lit-up' and 'bombed'.

Do we regard them as selves then, in the way we regard ourselves as selves?
What of a high level AI NPC in a game.

Now that is a more interesting idea, no, because the 'self' is an abstracted notion, the idea of the fractured cartesian theater is interesting. The self is a fiction, there are bodies.

Have I killed someone when I blast him away?

Nope, but people will respond as though a 'self' has passed.

I would have to say that if SHRDLU has the same "inner" experience of selfhood I do. And is aware of herself as a self, then I have must have more empathy toward her than a do a misquito.

But I know I'm dealing with an element of subjectivity here.

Ah, the heart of the matter, all we have is our idiomatic usage of symbols for communication.

And it seems that that's a very problematic and perhaps unwelcome thing to bring to the discussion.
The only content that can be analyzed are loops of self referencing circuitry.
What the heck could anyone mean by being a self otherwise?

The 'seer' behind the 'seeing', or some transcendent undefined blob of expectations and assumptions.

RIP SHRDLU. They pulled the plug on you!

 

[Dancing David]

No, it isn't. A Chinese mathematician using an entirely different set of symbols will do research entirely independently of his American counterpart, and come up with exactly the same results. That's because he's researching into things that are objectively true.

Not if one works in flat space (euclidean) and the other works in curved space (non-euclidean.)

 

[Dancing David]

It strikes me that actually the former is more of a test of awareness. Accessing memories of events seems to me less reliable. Being asked "Do you see the table" to me is inherently better here than being asked "Did you see the table yesterday?"

For me also, if SHRDLU has its full processing capacity engaged with the table then it cannot be aware of the table. There needs to be a sense of perspective, in humans usually generated by concurrent awareness of self or something else in the background. In Strong AI I imagine awareness is really just working memory.

In addition, for me I would need to know that SHRDLU possessed the programming to perform the feats it's being questioned about, or the programming to answer such questions correctly. Otherwise it could simply be answering in a pat way.



For me this is getting more complex now, because "aware" can have subtly different meanings in English. Do you mean "Do you know you're aware?" or "Are you aware of being aware?" The latter SHRDLU would certainly struggle with imo.



As I read Strong AI, SHRDLU will only have a subjective experience of selfhood if it has the capacity to construct selfhood. If SHRDLU has not the programming to teach itself what is SHRDLU, from the range of objects it is coming into contact with, then it will not have a "visual" sense of self. There are other aspects too.

Nick

So many interesting points to discuss. More later. memeory is a crucial part of language usage. No difference.