The Hard Problem of Gravity

[Paul C. Anagnostopoulos]

A 'hue' would be an element of 'seemingness' [yea, I made up a word ]. It stands to reason that, just like anything else, qualia can be broken down to simpler components. I chose to call such components 'hues' because it conveys the apparent continuity of subjective impressions; one quality of 'seemingness' can blend into another in a continuous manner.

Is a hue a quale, or have you introduced a new category of subjective experience?

~~ Paul

 

[Gate2501]

I don't understand why you would claim that lack of formal definition negates the reality of a phenomenon

I don't think that it would negate the reality of a phenomenon, but if the definition does not fit or coherently define the phenomenon in question then you might as well not even have a word for it.

It is sort of like saying, "hey, that basketball sure is fabblewuzzy!" and when someone asks you what "fabblewuzzy" means, you say that fabblewuzzy is subjective and can't be described, not by today's science at least, maybe in the future we will understand fabblewuzzy. You might as well have not said it in the first place.

Edit: I should add that it is ok to define things that we don't understand fully, so long as we define them by their BEHAVIOR.

 

[AkuManiMani]

um.... I was speaking of "hue" as we use it in speaking of color. Does your answer still apply?

Oh, I thought you were referring to my earlier use of the term 'hue' LOL

In answer to your question: Yes, it would still apply. I suppose since we humans are such visually oriented primates we tend to associate our sense of awareness the most with sight. But, generally speaking, qualitative experiences encompass all conscious sense impressions and thoughts -- whether they are strictly visual or not.

 

[AkuManiMani]

Is a hue a quale, or have you introduced a new category of subjective experience?

~~ Paul

Hehe. No its not a new category, per se, I'm essentially just splitting hairs

Here's the basic dichotomy I'm proposing:

quanta ['unit(s)' of information]
qualia ['hue(s)' of experience]

 

[AkuManiMani]

I don't think that it would negate the reality of a phenomenon, but if the definition does not fit or coherently define the phenomenon in question then you might as well not even have a word for it.

It is sort of like saying, "hey, that basketball sure is fabblewuzzy!" and when someone asks you what "fabblewuzzy" means, you say that fabblewuzzy is subjective and can't be described, not by today's science at least, maybe in the future we will understand fabblewuzzy. You might as well have not said it in the first place.

Edit: I should add that it is ok to define things that we don't understand fully, so long as we define them by their BEHAVIOR.

The difference, in this instance, is that the term being used is semantically well defined. 'Consciousness', by standard definition, encompasses every subjective experience. Such experiences are necessarily qualitative, hence the term qualia is used to refer to the range of all such experiences. The fact that it has not yet been scientifically defined does not make it any less real.

 

[rocketdodger]

This part I just don't follow you on.

Just a few hundred years ago human beings lacked many of the scientific definitions that we enjoy today. Their lack of formal definition didn't make events back then any less real. I don't understand why you would claim that lack of formal definition negates the reality of a phenomenon

They did not lack formal definitions.

Cave men had formal definitions of stuff.

A formal definition simply implies logical consistency.

A cave man could formally define "gravity" as "whatever makes stuff fall," -- there is no inconsistency there.

You could define "consciousness" in any number of logically consistent ways. The problem, of course, is that in doing so you are leaving open the possibility for things to be conscious that you don't want to say are conscious. That is why HPC proponents refuse to provide formal definitions.

 

[yy2bggggs]

Again, snipping some for length--if you think I snipped a crucial bit, let me know

Well, not a crucial bit. But we've had a pretty major disconnect. This part fascinates me though:

Out of curiosity, what wavelength of light are you using for unique red? And does "in context" mean fatiguing the opponent process blue channel? This is news to me unless you are meaning something quite different from what I am accustomed to.

Yes, most definitely, I am meaning something entirely different. I shine a laser pointer in a dark room. I have a single wavelength of light. I look at it, and perceive a spot on the wall. The percept I have about that spot is that it has the color red. That's it.

But your question had me confused--until I was able to relate why "unique red" and the blue opponent process would have to do with anything. You think I'm talking about a colorimetric primary, don't you?

In that case, I'm going to backtrack and move the other way. How do you produce said primary even with two colors of light? It's my understanding that there are no natural light conditions that can produce this primary. The colorimetric primaries are abstract.

"Said detector"... as nearly as I can see, said detector is defined circularly from the fact that we call things red.

That's not circular. That's referential.

Unless you are calling the entire person the "detector" (which does not appear to follow from what you have been writing), then there is every possibility that there are multiple detectors, one, or no detector at all, but only parallel processes only put into a unified whole at the level of the person.

I have a set of black boxes. They have a camera. When I put certain objects in view of their cameras, and press certain buttons, the black box produces certain outputs.

When I repeat the experiment, a funny thing happens. The same black box produces the same outputs for a set of objects. If I take the same set of objects and repeat the experiment with a different black box from the set, they also produce the same outputs. I find that I'm able to partition my objects based on the outputs produced by these black boxes. (Some objects produce inconsistent outputs... but a large number are consistent for all black boxes).

It's this ability of the black boxes to be used to partition these objects into sets, consistently, that I'm referring to. That suggests something about the black boxes, true. But it also suggests something else--it suggests that there's something common about those objects. They share something--something abstract. They share a capability to be partitioned into particular sets. The black boxes are detectors of this capability. And the black boxes have the capability to be affected by this abstract shared commonality. This is the only explanation I can imagine for why these black boxes are able to form meaningful partitions on the objects.

I'm calling the entire person the black box. Technically, sure, there's a remote possibility that there's no detector, but nothing worth entertaining (the categorizations I observed would have to be coincidental--the odds against it would be staggering); I don't think this is what you meant, so I'm going to go ahead and deny the likelihood of that possibility.

This also doesn't say anything about whether it's the black box as a whole, or a part within it, that "identifies" said commonalities. But I'll need a really good argument, in this case, to believe that the black box as a whole does it, and not being able to find some atomic part that is responsible for something I don't even believe is necessarily atomic in the first place... doesn't work.

The universality of color perception is not a done deal; certainly there are physiological constraints ...

Well, I'm not an expert in the field, so I'll tentatively take your word for it not being a done deal. It's certainly not a done deal for me--I'm not committed, just leaning. However, the paper you cited only confirmed a few suspicions I had. It was a very interesting read, but I'd be more interested in things that tear up my suspicions, which is the primary reason I'm suffering through this.

"Mechanical rather than merely learned" is a false dichotomy.

Your paper provided a better example:

There are, indeed, constraints on color categorization linked to the properties of the visual system. The most important constraint would be that similar items (as defined by perceptual discrimination) are universally grouped together. Thus, no language would exhibit categories that include two areas of color space but excludes an area between them.

So the categories here are culturally influenced, but there's an aspect to it that is not infinitely maleable. That's what's meant by the phrase (anything in the brain is going to be mechanical, be it innate, or learned--in this case, there are limits to the subjective categories based on contiguous areas in the color space--so the author claims).

The rest may be miscommunication. Refer to the black box perspective for a better description of what I'm inferring and why.

Note a particular nuance about my argument--I'm not opposed to "qualia" as a sort of "mass noun"; that is, a description of a subjective quality--i.e., "distinguishing characteristic"--of colors that are recognized. What I'm wary of is the identification of "quale"--singulars (like the "unique red" you brought up that you seemed to think I was arguing for) that are somehow "atomic components" of perception. (Still, though I'm wary of them, I don't rule them out).

 

[rocketdodger]

So, exactly what definition of information are you using which allows you to discard everything that's happening in the rock and thermostat except this one item which you happen to be interested in? Since you are so proud of your rigourous model, perhaps you could explain it. So far the language used has been entirely unscientific.

From wikipedia: http://en.wikipedia.org/wiki/Information

Information is the state of a system of interest

Lets walk through it in first order logic so you can see why a rock does not switch and a thermostat does.

We will be defining the predicate Switches() using the predicate State().

For all variables x and some constant C, State(x, x.STATE) iff x == C AND C == x.STATE.

In English, x is in state x.STATE if and only if there is some configuration C that matches x.STATE and x is in that configuration.

For all variables x, all variables y, and some constant S, Switches(S,x,y) implies State(y, y.ON) iff State(x, x.ON) AND State(S, S.ON).

In English, S switches x and y means that y is in the y.ON state if and only if both x and S are in their respective ON states as well.

Now, it is very clear that a functional thermostat satisfies the Switches relation, namely in an atomic sentence such as Switches(thermostat, external power line, air conditioning unit), where x and y have been replaced with constants.

It is not clear to me how a rock satisfies the Switches relation.

 

[rocketdodger]

The former implies the latter, though the latter doesn't necessarily imply the former. Experiencing a quality is a form of reasoning--in particular, it's tantamount to classifying something into an equivalence class.

Well, when you put it that way, yes, I agree. But I am not even close to convinced that any HPC proponent understands this.

Well, that's because quality is a noun, and used in this sense (at least how I parse the phrase), it doesn't specifically refer to perceptions (it's not the same in itself as "qualia"; it just mimics the form of the word--hey, I didn't come up with the phrase, I just claimed it made sense to me). "Quality of redness" is closer, but too ambiguous to reference perception still. But "experiencing the quality of redness" is pretty specific--that's referring in particular to the experiential aspect of the percept of red.

Go on to define "experiencing" then. You can't just use words without providing a formal definition!

Of course... if you just stuck with "reasoning" ... no explanatory power is lost and the icky definition process is already done and out of the way...

 

[yy2bggggs]

Go on to define "experiencing" then. You can't just use words without providing a formal definition!

Sure I can, else we can't even communicate!

Nevertheless, experiencing is the act of apprehension through your senses. So to tie the whole thing together, when you are "experiencing the quality of redness", you are apprehending the distinguishing characteristics of the color red using your senses.

In other words, "experiencing the quality of redness" is practicing a particular form of reasoning whereby you categorize something "visually" into an equivalence class.

Of course... if you just stuck with "reasoning" ... no explanatory power is lost and the icky definition process is already done and out of the way...

But I'd be wrong. The point isn't pain avoidance, it's error avoidance!

I can divide two three-digit numbers in my head (and often do, when filling up my gas tank). That involves reasoning, but it does not involve apprehending distinguishing characteristics of things through my senses (unless you extend your concept of "senses"). So the two concepts are not identical.

 

[AkuManiMani]

They did not lack formal definitions.

Cave men had formal definitions of stuff.

A formal definition simply implies logical consistency.

A cave man could formally define "gravity" as "whatever makes stuff fall," -- there is no inconsistency there.

You could define "consciousness" in any number of logically consistent ways. The problem, of course, is that in doing so you are leaving open the possibility for things to be conscious that you don't want to say are conscious. That is why HPC proponents refuse to provide formal definitions.

Eh? I'm not providing a formal definition because there is currently none to provide. I don't intend to make the same mistake of S-AI proponents of pretending I have one before we have scientifically gotten to the root of the problem.

The best we can do, at this point, is to start from instances that we know for certain have conscious experience [namely, us]. Via introspection, we can discern some of the general properties of our experience [like we did in the 'focus' thought experiment] and recognize that there are states in which our conscious experience is inactive. Internal, self-observation, is the first and most vital step in developing an understanding of consciousness.

From there, we can utilize our technology and the accumulated knowledge of biology, neuroscience, and physics to start to piece together what is physically going on in us during conscious experiences and generate plausible hypotheses. At this point, the general picture is still vague, but science is zeroing in on what consciousness is and what processes may give rise to it. Once we have our own consciousness figured out we can go about confidently proposing what other entities are conscious, in what ways, to what degrees and how we can seriously go about reproducing it from scratch.

Until there are more conclusive answers, dogmatically professing to have sufficient formal definitions of consciousness is intellectually irresponsible.

 

[Paul C. Anagnostopoulos]

Hehe. No its not a new category, per se, I'm essentially just splitting hairs.

Whether the hairs (qualia) can be split is precisely the question. Are qualia irreducible or not?

The difference, in this instance, is that the term being used is semantically well defined. 'Consciousness', by standard definition, encompasses every subjective experience. Such experiences are necessarily qualitative, hence the term qualia is used to refer to the range of all such experiences. The fact that it has not yet been scientifically defined does not make it any less real.

It doesn't make the experience less real, but it might make all the baggage associated with qualia disappear, just as the baggage associated with elan vital disappeared when we understood the chemical basis of life. The word quale might end up having been merely a placeholder for unobtained knowledge.

The sun still rises, but of course it does not.

~~ Paul

 

[AkuManiMani]

They did not lack formal definitions.

Cave men had formal definitions of stuff.

A formal definition simply implies logical consistency.

A cave man could formally define "gravity" as "whatever makes stuff fall," -- there is no inconsistency there.

Oh, and FYI, a formal definition of a thing is not the thing itself. The definitions utilized by cavemen were no more 'formal' than the colloquial definitions a layperson would use in day-to-day experience. Unless you want to argue that the universe didn't exist until humans developed language, formal mathematical systems, and modern science I don't think you wanna pursue this particular line of reasoning.

 

[AkuManiMani]

Whether the hairs (qualia) can be split is precisely the question. Are qualia irreducible or not?

That depends on what you mean by 'irreducible'.

Is information irreducible? Are 'elementary' particles irreducible? Does the question really matter in determining the gross properties of a phenomenon?

My guess is that qualia are emergent phenomena and that below a certain level of organization they do not exists as such.

It doesn't make the experience less real, but it might make all the baggage associated with qualia disappear, just as the baggage associated with elan vital disappeared when we understood the chemical basis of life. The word quale might end up having been merely a placeholder for unobtained knowledge.

The sun still rises, but of course it does not.

~~ Paul

I'd say that qualia have the same epistemological status that genes did back in Darwin's day or that memes currently do. Infact, I suspect that the actual physical elements of qualia are the same as those of memes.

From what bits and pieces I've been gleaming from some of the more cutting edge postulations in neuroscience it seems that mental elements, like memories, are information stored as distributed Fourier patterns across the nervous system. It stands to reason that since qualia [conscious sense impressions] and memes [replicable beliefs and behaviors] are stored as memories they would be encoded in such a holonomic system. Unlike genes, things like qualia, memes, and other mental elements are not composed of atoms and molecules, but the distributed patterns of information throughout one's brain and nervous system. Essentially, this view portrays the mind as a kind of holographic software entity encoded on the wetware of the brain. The 'ghost in the machine' is biological software.

Of course, these are just postulations and the details aren't fully worked out, but looking at the problem in such a way promises to bring a much greater understanding of the mind than was allowed by earlier conceptions.

 

[Paul C. Anagnostopoulos]

Is information irreducible? Are 'elementary' particles irreducible? Does the question really matter in determining the gross properties of a phenomenon?

No, but it does matter in determining whether the phenomenon is some sort of fundamental existent or whether it is explained in terms of simpler things. There are certainly many people who argue that consciousness is irreducible.

My guess is that qualia are emergent properties and that below a certain level of organization they do not exists as such.

If the term makes any sense at all, I agree that it is likely emergent. I think it will turn out that there is no crisp boundary between the conscious and nonconscious, and so the idea of a quale will become somewhat worthless.

Of course, these are just postulations and the details aren't fully worked out, but looking at the problem in such a way promises to bring a much greater understanding of the mind than was allowed by earlier conceptions.

In particular, the conception that the mind is some sort of separate entity that lives in the brain.

~~ Paul

 

[Mercutio]

Well, not a crucial bit. But we've had a pretty major disconnect. This part fascinates me though:

Yes, most definitely, I am meaning something entirely different. I shine a laser pointer in a dark room. I have a single wavelength of light. I look at it, and perceive a spot on the wall. The percept I have about that spot is that it has the color red. That's it.

But your question had me confused--until I was able to relate why "unique red" and the blue opponent process would have to do with anything. You think I'm talking about a colorimetric primary, don't you?

In that case, I'm going to backtrack and move the other way. How do you produce said primary even with two colors of light? It's my understanding that there are no natural light conditions that can produce this primary. The colorimetric primaries are abstract.

Just this bit for now--things to do.

I had to look up "colorimetric primary", and no, I am not speaking of that. I am thinking of psychophysics and color identification. It is possible (heck, it's easy) to ask people to describe colors in terms of percentages or mixes--or, for instance, to have them (in controlled conditions) manipulate a prism very carefully through a beam-splitter to produce, say, a light that looks like pure yellow (not yellow with a bit of green or yellow with a bit of red); this is "unique yellow". The same can be done with green (without yellow and without blue) or blue (without red or without green). One can also match the hue of one wavelength with a mixture of two others (virtually any two others, which is why the colorimetric primaries are an abstract). By the way, this is one way we measure spectral sensitivity curves.

What one cannot do is get a single wavelength that is perceived as unique red. Orange-red trails off into infrared before we see it as pure red, and violet-red trails into ultraviolet. We can, however, mix two wavelengths and get unique red--a red that appears to have no orange or violet to it.

This was all I was asking about--you claimed that we could get a "red" reaction with just one wavelength, and this was news to me. Your laser, though, will typically be an orange-red and not a unique red.

 

[westprog]

From wikipedia: http://en.wikipedia.org/wiki/Information



Lets walk through it in first order logic so you can see why a rock does not switch and a thermostat does.

We will be defining the predicate Switches() using the predicate State().

For all variables x and some constant C, State(x, x.STATE) iff x == C AND C == x.STATE.

In English, x is in state x.STATE if and only if there is some configuration C that matches x.STATE and x is in that configuration.

For all variables x, all variables y, and some constant S, Switches(S,x,y) implies State(y, y.ON) iff State(x, x.ON) AND State(S, S.ON).

In English, S switches x and y means that y is in the y.ON state if and only if both x and S are in their respective ON states as well.

Now, it is very clear that a functional thermostat satisfies the Switches relation, namely in an atomic sentence such as Switches(thermostat, external power line, air conditioning unit), where x and y have been replaced with constants.

It is not clear to me how a rock satisfies the Switches relation.

You could apply that mathematical description to things going on in the rock at almost any level. The rock itself expands and contracts due to temperature, and will inevitably impinge on something as it does so. Every atom is a switch.

The above mathematical description is perfect for the concept of a thermostat. It describes what we need a thermostat to do. But the physical thermostat will do an uncountable number of different things. If we want to describe them in informational terms, we could. It's just a matter of choice.

That's fine as long as we realise that it's an engineering description. In that case it's a perfectly reasonable way to look at it. It's even an OK description of the physical reality, if we accept that we are deliberately discarding most of what we know about the object.

But it is simply wrong to insist that the only bit of the thermocouple that we are interested in is the only bit that is doing anything - and that the rock isn't doing anything at all. And to insist that the thermocouple is a simple binary device, when if we look closely we'll find that current doesn't instantly stop or start, but increases and decreases depending on how tightly the contacts are touching.

There might be some kind of theory of Hard AI which accepts that the world is full of switches at every level, but that when they are all configured together in a particular way something will happen. I don't believe such theories, but they can't be totally ruled out. However, a theory which decides that only switches that we happen to be interested in have this effect is self-debunking, and serves only to demonstrate the limitations of the computational approach.

 

[westprog]

It doesn't make the experience less real, but it might make all the baggage associated with qualia disappear, just as the baggage associated with elan vital disappeared when we understood the chemical basis of life. The word quale might end up having been merely a placeholder for unobtained knowledge.

When a theory is found to explain qualia, then by all means, let the concept disappear with a better understanding. Perhaps it is just a placeholder for a better theory. But we don't have such a theory, and I don't see how we can make judgements based on a theory that doesn't exist.

 

[Paul C. Anagnostopoulos]

When a theory is found to explain qualia, then by all means, let the concept disappear with a better understanding. Perhaps it is just a placeholder for a better theory. But we don't have such a theory, and I don't see how we can make judgements based on a theory that doesn't exist.

I'm not sure tossing about the term quale in the meantime does us any good. In particular, it's clear that people have a spectrum of definitions in mind, all the way from some sort of irreducible dualistic notion to an acknowledged placeholder term for things we haven't nailed down yet. So then we have to argue about whether there really is a hard problem and such like.

But hell, it makes for fun threads.

~~ Paul

 

[westprog]

Just this bit for now--things to do.

I had to look up "colorimetric primary", and no, I am not speaking of that. I am thinking of psychophysics and color identification. It is possible (heck, it's easy) to ask people to describe colors in terms of percentages or mixes--or, for instance, to have them (in controlled conditions) manipulate a prism very carefully through a beam-splitter to produce, say, a light that looks like pure yellow (not yellow with a bit of green or yellow with a bit of red); this is "unique yellow". The same can be done with green (without yellow and without blue) or blue (without red or without green). One can also match the hue of one wavelength with a mixture of two others (virtually any two others, which is why the colorimetric primaries are an abstract). By the way, this is one way we measure spectral sensitivity curves.

What one cannot do is get a single wavelength that is perceived as unique red. Orange-red trails off into infrared before we see it as pure red, and violet-red trails into ultraviolet. We can, however, mix two wavelengths and get unique red--a red that appears to have no orange or violet to it.

This was all I was asking about--you claimed that we could get a "red" reaction with just one wavelength, and this was news to me. Your laser, though, will typically be an orange-red and not a unique red.

It's almost certainly the case that the qualia associated with "red" are strongly influenced by the nature of the chemical reactions taking place in the retina. It seems reasonable to presume that a creature that used different chemical reactions to perceive light would have an entirely different experience.

Our experience must surely also be coloured* by the associations of different colours. Blood and fire will inevitably give a different sensation to foliage. Hence a woman in a red dress will look different to one wearing green, on many levels.

*I put that in by accident and then decided to leave it. It's illustrative of the point.