Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Comment author: tadasdatys 21 July 2017 04:02:41PM 0 points [-]

Doing so does not buy you maximum precision in absolute terms

What sort of precision are you talking about? More generally, you're repeatedly said that the concept of consciousness is very useful. I don't think I've seen that usefulness. I suspect that elaborating here is your best bet to convince me of anything. Although even if you did convince me of the usefulness of the term, that wouldn't help the "robot pain" problem much.

You may believe that, but do you know it?

That's a slightly weird question. Is it somehow different from "why do you believe that" ? I believe it thanks to some rudimentary understanding of how brains and sensory organs work, and a lack of arguments to the contrary. It's very likely that "zapping" isn't quite sufficient, depending on how generously you interpret that word. But the idea that something cannot be learned through physical experiment, demands a lot of serious evidence, to say the least.

I'm not taking 1st person to mean 3rd person reports of 1st person experience.

If I can parse you correctly, you seem to be saying that a thought or memory is more true, in some sense, while stored in the brain, then if written down on paper. Obviously, paper is slower and less accurate. But you seem to be implying a more fundamental difference between those two methods of data storage. Why is that?

A realisation of type X has type X, a representation of type X has type "representation".

I like type theory. Let X be what I'm sitting on. Type of X is "chair", type of "chair" is "category", a painting of X is a representation of X, it is not a representation of "chair". Representations of "chair", in the same sense that painting represents X might not exist. Somehow I'm quite comfortable saying that an object of type Y is what represents Y. "Instantiates" might be the best word (curiously though, google uses "represent" to define it). Of course, the choice is quite arbitrary here. I don't see any confusion coming from it.

Comment author: TheAncientGeek 21 July 2017 04:17:01PM 0 points [-]

More generally, you're repeatedly said that the concept of consciousness is very useful

I have said that actual experience is useful to pin down the meaning s of words referring to exerpeince.

You may believe that, but do you know it?

That's a slightly weird question#

Not at all. That there is a difference betewen belief and knowledge is very standard.

I believe it thanks to some rudimentary understanding of how brains and sensory organs work, and a lack of arguments to the contrary.

There's an extensive literature of arguments to the contrary,

But the idea that something cannot be learned through physical experiment, demands a lot of serious evidence, to say the least.

It is the idea that you can learn aout the inward or 1st person by purely outward or 3rd person means that is contentious.

If I can parse you correctly, you seem to be saying that a thought or memory is more true, in some sense, while stored in the brain, then if written down on paper.

No, I am saying that my first person is me, and your first person is you. SO my first person information is my experience, not someone else's report of their experience.

Of course, the choice is quite arbitrary here. I don't see any confusion coming from it.

Well, you said that the two R words mean the same hting , when by established usage, they don't That looks like a source of confusion to me.

Comment author: tadasdatys 21 July 2017 04:02:44PM 0 points [-]

You are not treating the identity of pain with brain states as a falsifiable hypothesis.

No, I'm treating the identity of pain with the memories thoughts and behaviors that express pain, as unfalsifiable. In other words, I loosely define pain "the thing that makes you say ouch". That's how definitions work - the theory that the thing I'm sitting on is a chair is also unfalsifiable. At that point the identity of pain with brain states is in principle falsifiable, you just induce the same state in two brains and observe only one saying ouch. Obviously, there are various difficulties with that exact scheme, it's just a general sketch of how causality can be falsified.

There are uncontentious examples of multiply realisable things.

I don't recall suggesting that something isn't MR. I don't know why you think that MR is a problem for me. Like I said, there are multiple realizations of pain the same way that there are multiple realizations of chair.

Comment author: TheAncientGeek 21 July 2017 04:07:56PM *  0 points [-]

No, I'm treating the identity of pain with the memories thoughts and behaviors that express pain, as unfalsifiable.

Is that supposed to be a novel theory, or a dictionary definition?

I don't recall suggesting that something isn't MR

Your suggesting pain can't be instantiated in robots..

Comment author: tadasdatys 20 July 2017 05:07:06PM 0 points [-]

We do, on the other hand, know subjecively what pain feels like..

I know that the experience of stubbing my toe is called pain, and I know that what I'm sitting on is called a chair. But I don't know the "precise descriptions of the full gamut of atomic configurations which implement" them in either case. This is very normal.

You can't distingusih a brain-scan of someone seeing purple from a brainscan of someone tasting bitter.

You seem to be under impression that I advocate certain methods of examining brains over others. I don't know where you got that? I do believe that everything that could be learned about a brain could be learned by zapping one set of neurons and seeing if another set fires. But if it's more practical for you to show the brain something purple, and ask it to rate how bitter that felt, from 1 to 5, I have no problems with it. This method, while less direct, can be more useful (especially depending on the exact questions you want to answer). The problem, as I understand, is that you believe these two methods to be radically different, when they are not. It's as if you assume something is real, just because it comes out of people's mouths.

realisations or representations

I'm not assigning any different meanings to those words, at least not in this context. Are you? Interestingly, both words are pretty awkward to use here. And maybe I can agree that "realization" is a little better.

No one has made that argument.

Parts of my text are referring to the arguments I saw in wikipedia under "multiple realizaility". But the idea that Pain definitely is a thing that exists, rather than just a label for a set of things that your brain sometimes does, that are in some ways similar, is something I do find in your posts. In particular, you have to believe this to even ask whether robots feel pain.

and your reasons for sayign that are entirely non-empirical

I'm still waiting for your empirical reasons why "purple is not bitter", or better yet, "purple is not a chair", if you feel the concept of bitterness is too subjective.

Comment author: TheAncientGeek 21 July 2017 11:18:07AM *  0 points [-]

I'm not assigning any different meanings to those words, at least not in this context. Are you? Interestingly, both words are pretty awkward to use here. And maybe I can agree that "realization" is a little better.

The chair you are sitting on is a realisation; Van Gogh's painting of his chair at Arles is a representation. You can't sit on it.

But the idea that Pain definitely is a thing that exists, rather than just a label for a set of things that your brain sometimes does, that are in some ways similar, is something I do find in your posts. I

That's very vaguely phrased. There's are questions of whether pain has phenomenal qualities, whether it is totally reducible to physical behaviour, and whether it is multiply realisable. If pain doesn't have phenomenal properties, how do you decide which set of brain states get labelled as pain states?

But the concern is that you have no way of coming to know the answers to those questions. You have predetermined that everything must be treated as physics from the outset, so you will ineveitably get out the answer you put in. You are not treating the identity of pain with brain states as a falsifiable hypothesis.

There are uncontentious examples of multiply realisable things. Everything in computer science is MR - all algorithms, data structures , whatever. For the purposes of AI research, intelligence is assumed to be MR. There is no implication that MR things are things that "exist apart" from their realisations. So I don't know where you are getting that from.

In particular, you have to believe this to even ask whether robots feel pain.

I would have to believe pain is MR to believe that; but the objection cannot be that nothing is MR. You are apparently being inconsistent about MR.

"purple is not bitter

Colour and taste are different categories, therefore category error.

Comment author: tadasdatys 20 July 2017 05:07:06PM 0 points [-]

We do, on the other hand, know subjecively what pain feels like..

I know that the experience of stubbing my toe is called pain, and I know that what I'm sitting on is called a chair. But I don't know the "precise descriptions of the full gamut of atomic configurations which implement" them in either case. This is very normal.

You can't distingusih a brain-scan of someone seeing purple from a brainscan of someone tasting bitter.

You seem to be under impression that I advocate certain methods of examining brains over others. I don't know where you got that? I do believe that everything that could be learned about a brain could be learned by zapping one set of neurons and seeing if another set fires. But if it's more practical for you to show the brain something purple, and ask it to rate how bitter that felt, from 1 to 5, I have no problems with it. This method, while less direct, can be more useful (especially depending on the exact questions you want to answer). The problem, as I understand, is that you believe these two methods to be radically different, when they are not. It's as if you assume something is real, just because it comes out of people's mouths.

realisations or representations

I'm not assigning any different meanings to those words, at least not in this context. Are you? Interestingly, both words are pretty awkward to use here. And maybe I can agree that "realization" is a little better.

No one has made that argument.

Parts of my text are referring to the arguments I saw in wikipedia under "multiple realizaility". But the idea that Pain definitely is a thing that exists, rather than just a label for a set of things that your brain sometimes does, that are in some ways similar, is something I do find in your posts. In particular, you have to believe this to even ask whether robots feel pain.

and your reasons for sayign that are entirely non-empirical

I'm still waiting for your empirical reasons why "purple is not bitter", or better yet, "purple is not a chair", if you feel the concept of bitterness is too subjective.

Comment author: TheAncientGeek 21 July 2017 08:18:37AM *  1 point [-]

I know that the experience of stubbing my toe is called pain, and I know that what I'm sitting on is called a chair. But I don't know the "precise descriptions of the full gamut of atomic configurations which implement" them in either case. This is very normal.

But not much of an argument for using semantics grounded in (physical) reality. Doing so does not buy you maximum precision in absolute terms, and , what is worse, the alternative , of grounding terms for types of experience is 1st person experience, can give you more precision.

You seem to be under impression that I advocate certain methods of examining brains over others. I don't know where you got that? I do believe that everything that could be learned about a brain could be learned by zapping one set of neurons and seeing if another set fires.

You may believe that, but do you know it?

But if it's more practical for you to show the brain something purple, and ask it to rate how bitter that felt, from 1 to 5, I have no problems with it. This method, while less direct, can be more useful (especially depending on the exact questions you want to answer). The problem, as I understand, is that you believe these two methods to be radically different, when they are not.

The difference is that I accept the possibility that first person evidence could falsify 3rd person theory.

It's as if you assume something is real, just because it comes out of people's mouths.

I'm not taking 1st person to mean 3rd person reports of (someone elses) 1st person experience.

Comment author: tadasdatys 20 July 2017 04:26:42PM 0 points [-]

How do you know?

Well, for one, you have been unwilling to share any such knowledge. Is it a secret, perhaps?

https://en.wikipedia.org/wiki/Global_Workspace_Theory

I see a model that claims to reproduce some of the behaviors of the human mind. Why is that relevant? Where are your subjective experiences in it?

Also, to clarify, when I say "you know nothing", I'm not asking for some complex model or theory, I'm asking for the starting point from which those models and theories were constructed.

prove that it is a stupid question.

Proof is a high bar, and I don't know how to reach it. You could teach me by showing a proof, for example, that "is purple bitter" is a stupid question. Although I suspect that I would find your proof circular.

Comment author: TheAncientGeek 20 July 2017 05:56:48PM *  1 point [-]

Well, for one, you have been unwilling to share any such knowledge. Is it a secret, perhaps?

It's very difficult to prove that something is impossible, and you can't do it by noting that it has never happened yet.

Where are your subjective experiences in it?

I was responding to your claim that "there is nothing that you know about consciousness, from which you can derive a more accurate and more material description.". This has been done, so that claim was false. You have shifted the ground.

that "is purple bitter" is a stupid question.

Purple is a colour, bitter is taste, therefore category error.

Proof is a high bar

Then why be so sure about things? Why not say "dunno" to "can robots feel pain?".

Comment author: tadasdatys 20 July 2017 04:09:46PM 0 points [-]

Not if "perfectly good" means "known".

It's ok, it doesn't. Why do people keep bringing up current knowledge?

Comment author: TheAncientGeek 20 July 2017 04:14:14PM *  1 point [-]

Because we are trying to communicate now, but your semantic scheme requires knowledge that is only available in the future , if at all.

Comment author: tadasdatys 20 July 2017 06:35:03AM 0 points [-]

That's the sense that matters.

That's not a very interesting sense. Is height also subjective, since we are not equally tall? This sense is also very far from the magical "subjective experience" you've used. I guess the problematic word in that phrase is "experience", not "subjective"?

Comment author: TheAncientGeek 20 July 2017 04:10:46PM *  0 points [-]

Height is not a subjective judgement because it is not a judgement. If judgements are going to vary, that matters, because then who knows what the truth is?

Comment author: tadasdatys 19 July 2017 05:31:57PM 0 points [-]

having a first person perspective

What does "not having a first person perspective" look like?

gerrymnadered definition of meaning

I find my definition of meaning (of statements) very natural. Do you want to offer a better one?

subjectivity

I think you use that word as equivalent to consciousness, not as a property that consciousness has.

I can ask and answer subjective questions of myself, like how do I feel, what can I remember, how much do I enjoy a taste.

All of these things have perfectly good physical representations. All of them can be done by a fairly simple bot. I don't think that's what you mean by consciousness.

Comment author: TheAncientGeek 20 July 2017 03:18:08PM *  1 point [-]

All of these things have perfectly good physical representations.

Not if "perfectly good" means "known".

Comment author: tadasdatys 20 July 2017 06:49:12AM 0 points [-]

It is possible to use language meaningfully without knowing exactly how it pans out in terms of precise configurations of matter

I have never claimed otherwise. In fact, there is literally nothing that I have exact description of, in terms of matter - neither pain nor chairs. But you have to know something. I know that "chair is what I sit on" and from that there is a natural way to derive many statements about chairs. I know that "gravity is what makes things fall down", and from that there is a fairly straightforward way to the current modern understanding of gravity. There is nothing that you know about consciousness, from which you can derive a more accurate and more material description.

Treating pain semantically as some specific brain activity buys you nothing

It buys me the ability to look at "do robots feel pain" and see that it's a stupid question.

And without committing yourself to evil dualism.

What evil dualism?

Comment author: TheAncientGeek 20 July 2017 02:10:38PM 0 points [-]

There is nothing that you know about consciousness, from which you can derive a more accurate and more material description.

How do you know? And what of things like https://en.wikipedia.org/wiki/Global_Workspace_Theory ?

It buys me the ability to look at "do robots feel pain" and see that it's a stupid question.

It doesn't seem to have given you the ability to prove that it is a stupid question.

Comment author: tadasdatys 20 July 2017 08:19:19AM 0 points [-]

But no one knows precise descriptions of the full gamut of atomic configurations which implement pain.

Sure, but what does that have to do with anything? Does "objective" mean "well understood" to you?

multiple realisability

There are multiple representations of pain the same way that there are multiple representations of chair.

It is ridiculous how much of this debate is about the basic problem of classification, rather than anything to do with brains. Flawed reasoning starts with a postulate that "Pain" exists and then asks, what physical states correspond to it. And when told that "pain is the activity in region X", it somehow feels that "activity in Y could also be described as pain", is a counter argument. Good reasoning starts with noticing that people say "ouch" when they stub their toes, or that subbing a toe has a very distinct feeling, and then asks, what causes/predicts these actions/differences, and then wonders, how could we best classify these.

your subjective intuitions

Calling my reasoning, even if not fully formal, "subjective intuitions" seems rude. I'm not sure if there is some point you're trying to express with that.

You accept multiple realisability for intelligence, but not for consciousness. That is arbitrary.

Not sure where you see me talking about intelligence. But intelligence is far more well defined and measurable than consciousness. Multiple realizability has nothing to do with that.

Comment author: TheAncientGeek 20 July 2017 11:49:48AM *  2 points [-]

But no one knows precise descriptions of the full gamut of atomic configurations which implement pain.

Sure, but what does that have to do with anything?

We do, on the other hand, know subjecively what pain feels like..

Does "objective" mean "well understood" to you?

That's not the point. The point is that if we have words referring to subjective sensations, like "purple" and "bitter", we can distinguish them subjectively. But if we discard out subjective insight into them, as you are proposing, and replace them with vague objective descriptions -- vague, because no one knows precise descriptions of the full gamut of atomic configurations which implement pain. -- then you take a step backwards. You can't distingusih a brain-scan of someone seeing purple from a brainscan of someone tasting bitter. Basing semantics on objective facts, or "reality" as you call it. only works if you know which fact is which. You are promoting something which sounds good, but doesn't work -- as a research program. Of course it works just fine at getting applause from an audience of dualism-haters.

.multiple realisability

There are multiple representations

Are you talking about realisations or representations?

Flawed reasoning starts with a postulate that "Pain" exists and then asks, what physical states correspond to it. And when told that "pain is the activity in region X", it somehow feels that "activity in Y could also be described as pain", is a counter argument.

No one has made that argument. The point is not that it is not ultimately true that subjetive states are brain states, it is that rejecting the subjective entirely, at this stage, is not useful. Quite the reverse. Consciousness is the only thing we know from the inside--why throw that away?

Good reasoning starts with noticing that people say "ouch" when they stub their toes, or that subbing a toe has a very distinct feeling, and then asks, what causes/predicts these actions/differences, and then wonders, how could we best classify these.

If we know what casues or predicts something, then we can sometimes reproduce it in radically different ways. That is the basis of artificial intelligence: realising intelligence in a medium other than a brain.

But you are saying that pain cannot be realised by a robot, and your reasons for sayign that are entirely non-empirical.

We might be able to refine the concept of consciousness as part of a research programme, but research programmes have to start with folk concepts.

View more: Next