My guess is that you are the victim of a failure of imagination: specifically, you fail to imagine everything an imperfect information-processing model-maintaining agent might falsely believe. Specifically, such an agent might falsely believe that aspects of its own operation are irreducible primitives. You keep on asserting that subjective experiences are irreducible to the primitives of the standard physical model, but you have not presented anything that I consider evidence for that.
The only way I can think of for you to make progress on moving me towards your point of view is for you to point out a problem with the standard model of physics using the vocabulary of the technical theory of how any agent can come to have an accurate model of its environment (Jaynes, Pearl, universal prior, Solomonoff induction, etc).
You keep on asserting that subjective experiences are irreducible to the primitives of the standard physical model, but you have not presented anything that I consider evidence for that.
Whereas from my perspective, no-one is explaining how any particular higher-level physical property can be identified with a color—for example. As I just asked Robin Z, please explain to me what’s green about a causal disposition or a physical motion.
I have yet to see any such explanation. Instead I just see assertion of identity, or a discourse structured to avoid talking directly about color.
I am skeptical that quantitative epistemology is much use here, because it is usually practised in a mindset which is already treating everything abstractly.
Perhaps the key is to get people into a state of mind in which they are genuinely attending to the “qualia” themselves, and in which it is not assumed that they must be reducible to the physics we have, and then to have them ponder afresh whether the alleged identities above (green as a causal disposition, green as a physical motion) actually make sense. Also throw in a warning to beware treating the possibility of systematic association as identity: imagining that the occurrence of greenness is always accompanied by some physical process or condition, is not the same thing as perceiving that greenness could be identical to the physical counterpart.
As I just asked Robin Z, please explain to me what’s green about a causal disposition or a physical motion.
And I’ve answeredyouseveral times, which is why it’s specifically my comments that you avoid.
Let’s go over this again: computer programs are in the very same dilemma. They use generated symbols. GensymA refers to this data. GensymB refers to that data. MetaGensym1 refers to the group {GensymA, GensymB, …}.
It can tell any two gensyms apart. It can tell any metagensym group apart. But from the program’s perspective, it cannot tell what is “GensymA-ish” about this data, or “GensymB-ish” about that data—just whether they are or aren’t. Between two program instances, all of this (within limits) could be switched around, and there would be no multi-program GensymA.
You already know how this situation arises from the physicalist reductionist account.
You simply have to recognize yourself as being in that same scenario. Your internal, truly-part-of-you labels for different phenomena are the qualia—which accounts for the problematic aspects of qualia.
Does this resolve the issue completely? Of course not. Among many other things, we need to figure out what (seemingly efficient) data representation method the brain uses that causes the specific aspects of color, like its ability to vary in shade, and vary orthogonally to the sounds you hear. But there’s a clear research program there and a coherent picture from a reductionist account.
Incidentally, it may amuse you that others have (independently, for all I can tell) come to entirely similar conclusions:
But nothing requires us to make such an invocation. We don’t have to know how we identify or re-identify or gain access to such internal response types in order to be able so to identify them. This is a point that was forcefully made by the pioneer functionalists and materialists, and has never been rebutted (Farrell, 1950, Smart, 1959). The properties of the “thing experienced” are not to be confused with the properties of the event that realizes the experiencing. To put the matter vividly, the physical difference between someone’s imagining a purple cow and imagining a green cow might be nothing more than the presence or absence of a particular zero or one in one of the brain’s “registers”. Such a brute physical presence is all that it would take to anchor the sorts of dispositional differences between imagining a purple cow and imagining a green cow that could then flow, causally, from that “intrinsic” fact. (I doubt that this is what the friends of qualia have had in mind when they have insisted that qualia are intrinsic properties.)
your phenomenal experience of blue, is your label for the kind of data that is blue
And I presume that the “blue data” is called blue, not because it is literally blue in the old-fashioned sense, but because it’s caused by physical “blueness”, or just because that’s the name we’re using for a particular range of data values.
To paraphrase what I just said to Richard: in effect, you are saying that the experience of color is the experience of colorlessness, plus a color label. Which is the same as saying that I don’t actually see color, I just think I see color.
This is getting tiresome. The link you just gave was actually to a time when I was summarizing my previous attempt to correct a further previous misunderstanding of my position.
You haven’t actually addressed any of it, including here. I gave a specific example of this model of qualia in a different context: generated symbols in programing. Instead of just an inflammatory strawman plus a non-sequitur,
you are saying that the experience of color is the experience of colorlessness, plus a color label. Which is the same as saying that I don’t actually see color, I just think I see color.
how about you actually say where you think that model breaks down?
How about you stop repeating the same confusion between the linguistic label and the computational label, like you’re doing here:
And I presume that the “blue data” is called blue, … because that’s the name we’re using for a particular range of data values.
Again (this is at least the third time I’ve explained this to you): the phenomenal experience of blue is not the same thing as the name “blue”. You experience seeing blue whether or not you have the term “blue” (or “azul” or “blau” or “bleu” or “aoi”). Rather, the phenomenal experience of blue is what it is like to be a program that has classified incoming data as being a certain kind of light, under the constraint of having to coherently represent all of its other data (other colors, other visual qualities, other senses, other combined extrapolations from multiple senses, etc) but with limited comparison abilities.
Yes, as part of your use of a language, you can assign the label “blue”. But that’s not how I’m explaining phenomenal blue. I’m explaining phenomenal blue as your architecture’s direct label for a kind of light, below the level at which you can see it work. To experience blue is to feel your cognitive architecture assigning a label to sensory data.
Now, you may have a reason to reject this approach. You may have reason to believe that the associated research program will turn out surprisingly fruitless. You may have an alternative which looks more promising.
But I have no way of knowing that, when the only response that you give to the reasoning that I’ve just given (and have given in some form or another five times now) is to ignore it or respond to a mischaracterization of it.
Now, try again, and this time, communicate to me what caused you to reject this approach at the time you considered it during the last eight years. I would love to know what you uncovered in trying this out.
To move me toward your point of view, you would need to do one of three things. (1) Show me that I am wrong in my expectation that your proposal will lead to an “ontology” (an account of reality) with significantly higher Kolmogorov complexity than the conventional ontology. (2) Present evidence that outweighs the higher Kolmogorov complexity. In particular, present evidence that not only prefers an ontology consistent with your proposal, but also does so to such a degree so as to outweigh the higher Kolmogorov complexity. (3) Cause me to come to doubt the epistemology I am using (universal prior, Bayesian updating, etc).
It would interest me to know whether you find any fault in my position as expressed above, Mitchell, because your finding a fault would be a strong sign that our differences in this thread stem from differences over epistemology.
Perhaps the key is to get people into a state of mind in which . . . it is not assumed that [qualia] must be reducible to the physics we have
I have not assumed anything of the sort. I am simply noticing that in contrast to what you seem to believe, qualia are not sufficiently strong evidence against the conventional ontology to satisfy condition (2) above.
(I will now quote again from the same sentence I quoted from above, but this time I will omit a different passage.)
Perhaps the key is to get people into a state of mind in which they are genuinely attending to the “qualia” themselves . . .
Do you sincerely believe that the people replying to you here neglected genuinely to attend to “the qualia themselves” when they considered your words and how to reply to them? I assure you that I for one did not. Just now, in fact, I caused myself to experience blueness while reflecting on your argument. It was no more persuasive than the last couple of times I did it.
I caused myself to experience blueness while reflecting on your argument.
Did you remind yourself that what you are experiencing is inside your head, which according to conventional physics is composed entirely of colorless entities, and notice that nonetheless, something inside your head—a particular sensation—managed to be blue? If so, how did you deal with the contradiction?
We can have a dispute about the Kolmogorov complexity of different explanations once we agree on what it is that we’re trying to explain.
Did you remind yourself that what you are experiencing is inside your head, which according to conventional physics is composed entirely of colorless entities, and notice that nonetheless, something inside your head—a particular sensation—managed to be blue? If so, how did you deal with the contradiction?
If I find that I am able to experience blueness and to experience redness, then my brain must have at least two states, one that corresponds to the blue experience and one that corresponds to the red experience, or we have a contradiction.
The state of my brain that corresponds with the blue experience can be a normal, ordinary, conventional physical state. You have made no progress in persuading me—or as far as I can tell anyone else who has commented on Less Wrong—that it must be a special state where a special state is defined as a state that cannot be modeled by the conventional ontological model.
The thing that you refer to as a contradiction is only a contradiction if one mistakenly clings to a particular causal model (or a particular set of causal models) of the sensation of blueness.
I am using “state” the way the computer scientists use it, namely, to mean a configuration of reality or of an “identifiable” aspect of reality (such as my brain) that can change as a function of time.
Well, thanks for replying to my questions.
My guess is that you are the victim of a failure of imagination: specifically, you fail to imagine everything an imperfect information-processing model-maintaining agent might falsely believe. Specifically, such an agent might falsely believe that aspects of its own operation are irreducible primitives. You keep on asserting that subjective experiences are irreducible to the primitives of the standard physical model, but you have not presented anything that I consider evidence for that.
The only way I can think of for you to make progress on moving me towards your point of view is for you to point out a problem with the standard model of physics using the vocabulary of the technical theory of how any agent can come to have an accurate model of its environment (Jaynes, Pearl, universal prior, Solomonoff induction, etc).
Whereas from my perspective, no-one is explaining how any particular higher-level physical property can be identified with a color—for example. As I just asked Robin Z, please explain to me what’s green about a causal disposition or a physical motion.
I have yet to see any such explanation. Instead I just see assertion of identity, or a discourse structured to avoid talking directly about color.
I am skeptical that quantitative epistemology is much use here, because it is usually practised in a mindset which is already treating everything abstractly.
Perhaps the key is to get people into a state of mind in which they are genuinely attending to the “qualia” themselves, and in which it is not assumed that they must be reducible to the physics we have, and then to have them ponder afresh whether the alleged identities above (green as a causal disposition, green as a physical motion) actually make sense. Also throw in a warning to beware treating the possibility of systematic association as identity: imagining that the occurrence of greenness is always accompanied by some physical process or condition, is not the same thing as perceiving that greenness could be identical to the physical counterpart.
And I’ve answered you several times, which is why it’s specifically my comments that you avoid.
Let’s go over this again: computer programs are in the very same dilemma. They use generated symbols. GensymA refers to this data. GensymB refers to that data. MetaGensym1 refers to the group {GensymA, GensymB, …}.
It can tell any two gensyms apart. It can tell any metagensym group apart. But from the program’s perspective, it cannot tell what is “GensymA-ish” about this data, or “GensymB-ish” about that data—just whether they are or aren’t. Between two program instances, all of this (within limits) could be switched around, and there would be no multi-program GensymA.
You already know how this situation arises from the physicalist reductionist account.
You simply have to recognize yourself as being in that same scenario. Your internal, truly-part-of-you labels for different phenomena are the qualia—which accounts for the problematic aspects of qualia.
Does this resolve the issue completely? Of course not. Among many other things, we need to figure out what (seemingly efficient) data representation method the brain uses that causes the specific aspects of color, like its ability to vary in shade, and vary orthogonally to the sounds you hear. But there’s a clear research program there and a coherent picture from a reductionist account.
Incidentally, it may amuse you that others have (independently, for all I can tell) come to entirely similar conclusions:
The above comes from Quining Qualia by Daniel Dennett—the citations are to:
Farrell (1950). “Experience,” Mind, 59, pp.170-98.
Smart, J.C. (1959). “Sensations and Brain Processes,” Philosophical Review, LXVIII, pp.141-56.
You state the essence of your view here:
And I presume that the “blue data” is called blue, not because it is literally blue in the old-fashioned sense, but because it’s caused by physical “blueness”, or just because that’s the name we’re using for a particular range of data values.
To paraphrase what I just said to Richard: in effect, you are saying that the experience of color is the experience of colorlessness, plus a color label. Which is the same as saying that I don’t actually see color, I just think I see color.
This is getting tiresome. The link you just gave was actually to a time when I was summarizing my previous attempt to correct a further previous misunderstanding of my position.
You haven’t actually addressed any of it, including here. I gave a specific example of this model of qualia in a different context: generated symbols in programing. Instead of just an inflammatory strawman plus a non-sequitur,
how about you actually say where you think that model breaks down?
How about you stop repeating the same confusion between the linguistic label and the computational label, like you’re doing here:
Again (this is at least the third time I’ve explained this to you): the phenomenal experience of blue is not the same thing as the name “blue”. You experience seeing blue whether or not you have the term “blue” (or “azul” or “blau” or “bleu” or “aoi”). Rather, the phenomenal experience of blue is what it is like to be a program that has classified incoming data as being a certain kind of light, under the constraint of having to coherently represent all of its other data (other colors, other visual qualities, other senses, other combined extrapolations from multiple senses, etc) but with limited comparison abilities.
Yes, as part of your use of a language, you can assign the label “blue”. But that’s not how I’m explaining phenomenal blue. I’m explaining phenomenal blue as your architecture’s direct label for a kind of light, below the level at which you can see it work. To experience blue is to feel your cognitive architecture assigning a label to sensory data.
Now, you may have a reason to reject this approach. You may have reason to believe that the associated research program will turn out surprisingly fruitless. You may have an alternative which looks more promising.
But I have no way of knowing that, when the only response that you give to the reasoning that I’ve just given (and have given in some form or another five times now) is to ignore it or respond to a mischaracterization of it.
Now, try again, and this time, communicate to me what caused you to reject this approach at the time you considered it during the last eight years. I would love to know what you uncovered in trying this out.
I have aggregated my latest responses here.
What do you think you do when you see, dance a minuet to the music of Bach? Seeing is an act of thought.
To move me toward your point of view, you would need to do one of three things. (1) Show me that I am wrong in my expectation that your proposal will lead to an “ontology” (an account of reality) with significantly higher Kolmogorov complexity than the conventional ontology. (2) Present evidence that outweighs the higher Kolmogorov complexity. In particular, present evidence that not only prefers an ontology consistent with your proposal, but also does so to such a degree so as to outweigh the higher Kolmogorov complexity. (3) Cause me to come to doubt the epistemology I am using (universal prior, Bayesian updating, etc).
It would interest me to know whether you find any fault in my position as expressed above, Mitchell, because your finding a fault would be a strong sign that our differences in this thread stem from differences over epistemology.
I have not assumed anything of the sort. I am simply noticing that in contrast to what you seem to believe, qualia are not sufficiently strong evidence against the conventional ontology to satisfy condition (2) above.
(I will now quote again from the same sentence I quoted from above, but this time I will omit a different passage.)
Do you sincerely believe that the people replying to you here neglected genuinely to attend to “the qualia themselves” when they considered your words and how to reply to them? I assure you that I for one did not. Just now, in fact, I caused myself to experience blueness while reflecting on your argument. It was no more persuasive than the last couple of times I did it.
Did you remind yourself that what you are experiencing is inside your head, which according to conventional physics is composed entirely of colorless entities, and notice that nonetheless, something inside your head—a particular sensation—managed to be blue? If so, how did you deal with the contradiction?
We can have a dispute about the Kolmogorov complexity of different explanations once we agree on what it is that we’re trying to explain.
If I find that I am able to experience blueness and to experience redness, then my brain must have at least two states, one that corresponds to the blue experience and one that corresponds to the red experience, or we have a contradiction.
The state of my brain that corresponds with the blue experience can be a normal, ordinary, conventional physical state. You have made no progress in persuading me—or as far as I can tell anyone else who has commented on Less Wrong—that it must be a special state where a special state is defined as a state that cannot be modeled by the conventional ontological model.
The thing that you refer to as a contradiction is only a contradiction if one mistakenly clings to a particular causal model (or a particular set of causal models) of the sensation of blueness.
I am using “state” the way the computer scientists use it, namely, to mean a configuration of reality or of an “identifiable” aspect of reality (such as my brain) that can change as a function of time.
I have aggregated my latest responses here.
And a color blind person may never see green. I wonder, if everybody was color blind would green cease to be a fundamental property of physics?
Back up: since when has color been a fundamental property of physics? Wavelength is, but wavelength and color are not identical.
Some day I’m going to learn not to speak in facetious riddles.
Why not? Roughly speaking, if everyone was 2-dimensional, why would we see the world as 3-dimensional?
Edit: this came out wrong, for obscure reasons.
Are you substituting in “believed to be a fundamental property of physics” for “being a fundamental property of physics”?
My fault. That comment was some kind of cached thought that I can’t quite examine now to explain what I thought I was writing.