The War of Qualia and Consciousness, Battle 1

This discussion set has happened on The Brights.Net forums Mostly between a gent using the handle "Emo" and myself, who has kindly allowed me to use the excerpts here. The discussion began in the thread Why a Super might support the Brights in which I posted my description of our mind's mentation as relates to consciousness, From Beast to Brightman, An Opera of the Evolution of Consciousness in Five Acts and continued in the thread The Hard Problem of Hunting the Wild Quale also appearing here.

The question before the house is does this answer "The Hard Problem" as elaborated in http://consc.net/papers/facing.html . David Chalmers' "Facing the Hard Problem".

My answer is yes. And I'm curious to see whether others can shoot it down.

"The Hard Problem" is

And the answer is simply that the experience of the self is the projection of oneself in regard to these things (in room 2c, in my model) cotemporaneous with memories of excerpts and abstracts of the real past sensory experience (including possible inter body sensations) in pursuit of whatever goal the decision process in the brain is seeking. Including possibly answering the question about "do I have these qualia?", In which case it having previously conjured the projection of its puppet of self with those past experiences and they and associated relations coming down through the decision mechanism, it projects itself as the puppet nodding sagely, yes I know those qualia, and in consequence typing this paragraph (after running thousands of little projections about just how having such qualia relate to the abstract model of consciousness referenced and building anticipatory schemas of doing so). "Why is it that when our cognitive systems engage in visual and auditory information-processing, we have visual or auditory experience: the quality of deep blue, the sensation of middle C?" Very simply because "when our cognitive systems engage in visual and auditory information-processing" it explicitly means referencing memories of excerpts and past abstractions of those past real experiences. We seem to have a similar experience to the real experience because the information passes through with reference to those real experiences, but it is not as rich and full as the real experience, for two reasons, the mind simply lacks the resources for full storage of the whole experience when it first occurred and doesn't have it to pass through, just the abstracts, and also the data is tagged as being part of a simulation. The mind knows it is not "experiencing" the sensory experiences live.

But there's an additional and much subtler problem here. You see, I never had a quale until I understood qualia.

Of course I had the experience of seeing red (to revert to the standard example) and would recognise red in any image falling across my retinas. And the word "red" and all its variants (magenta, ochre, brick) would trigger hundreds of past experiences of which one abstracted part was the experience we commonly mean by red. But I never had the "quale" of redness ... depending on where you define it.

But there's no mystery to having and knowing the experience of red. We have physical experiences. We know our brains cut them up, store and recombine them in pieces and can summon those memories up along with their sundry associations. Hardly a hard problem conceptually, though we're far from understanding the mechanics in our brains, our computers and cameras can do the same thing trivially, even though they have no conception of what they're doing while doing it.

So what is a quale? So I learn about qualia. In the puppet theater the self stands with the question, "do you know about abstractions, boy?", and back down through the C wing of levels (let's call them "scenes" from now on) 2 through 4 come the associations of the words "abstractions" and "you", and 4c directs 2c to have the puppet nod wisely. "So boy, do you understand the abstraction 'What x is like?'" ... and again the puppet nods. "So boy, a quale is the 'what it is like' for something we experience. Do you get it?" Happily the puppet nods, relieved that he wasn't about to be scolded for his obtuseness. "So explain to me the quale of red!" shouts the phantom professor in Descartes' theater. The puppet looks around and panics, then finally stammers "Um.... Er... I thought I understood, but I don't think I can ... there's just no way ..." "Excellent!", shouts the imaginary professor, "You've got it!".

But alone in the dark Pinocchio untangles his strings. Where did the mystery come from? There is no mystery to what red is. There is no real mystery to what our brains are doing in sorting out red as we can have our machines do it. And there is no mystery, conceptually, as to how it's done for the same reason though we don't yet have a good correlation between our machines and the mechanisms in goo. At first glance "qualia" like pain and hunger may seem to be candidate mysteries, but if the brain can create an experience out of light, there's no reason it should have a problem doing so about more overt chemical or physical problems. So where is the mystery? And the answer I see is that it was added after the fact.

The problem of the quale of red is nothing about how the brain/mind sees red, it is in seeing red as an abstraction of the experience of red, not the experience itself, nor in the myriad ways memories of such experiences may be recalled in the mind.

The problem, my dear professor, lies not with ourselves, but with our philosophies.

-- TWZ

Exactly. The whole thing works like that, start to end. There isn't a hint of consciouness until the puppet takes the stage in 2c of Act 4.

The "hard problem" is simply the brain mouthing through the puppet in 2c that it knows it has had the experience of any particular quale and has the knowledge that it will recognise it again.

-- TWZ

If your autoland system had a model of itself as an actor in the play, say it "knows" that it can't handle more than x planes coming in under various conditions, and modeled various scenarios which necessarily included its own assessment of its capabilities using its own historical data of its own behavior and reactions, then yes it would be conscious, though in a fairly limited domain.

-- TWZ

Corrective addendum. As I see it from my model, conscious and consciousness are two different things, which I've long known putting this model together, but neglected to make note of the strong difference in my little opera.

A critter becomes "conscious" somewhere about Act 2. If it has anticipatory data from past experience that it uses to compare with present data it is "conscious" (though it lacks a means to tell anyone about it). In a very limited sense your autoland system probably has limited "conscious" but no consciousness ... it likely, from what little I know about such, has anticipated glide paths for various aircraft and various conditions and is "conscious" of the variance and acts accordingly. The difference is slight as a conscious animal mostly generates the anticipation from past experience, whereas you've undoubtedly programmed the ideal glidepath in from historical and ideal path data outside of ongoing system behavior.

But it never has "consciousness" until and unless it has a model of itself which is significant and adaptive to its goals in the action models it generates.

-- TWZ

Another poster, Myron, points us another of Chalmers' papers, Consciousness and its Place in Nature

Meanwhile EMO, addressing another post has posited building an extremely sophisticated aircraft landing system he calls "GALS".

Emo, you're a mystic? (mystery and mystic have the same cognate).

There's a hypothesis on the table. You need to displace it with a better one or falsify it by demonstrating an obviously true property, relationship, or consequent of consciousness directly contradicting that model. I'll be the first to grant that "my" model is probably primitive and full of holes, but we have to start somewhere. Aristotle may have been dead wrong about gravity (as I'll admit it's quite possible that I may be) , but his "model" gave Galileo a basis for comparison.

In re Chalmers, the paper needs more study but in a quck scan through ...

I'm a materialist-b with perverse reductionist leanings.

In our universe, zombies are not possible: if they are constructed atom for atom (and energy states within reason - no set of 1.21GeV electrons blowing the whole thing sky high...) then they will have consciousness and qualia exactly as we do. Zombies are "conceivable" ... as is an ontological entity with property A and not A, but, if a materialistic model of consciousness is correct, are neither metaphysically nor materially possible. (the material model negates the logical possibility as well as being a counter-demonstration to the physical possibility).

If I take an Easter egg into Mary's room does she know what color it is? Mary lacks "knowledge" that every two year old has. On the other hand if she knows how her eyes and brain are connected, and how her brain *would* react to seeing different colors ... and has a real-time readout of what is happening in her brain which she can read and interpret as she looks at the egg, then she *does* have the knowledge and does know (or can figure out) what color the egg is. If a material model of consciousness is correct, the phenomenology of a real sensory experience is nothing but the reaction of the brain in various complex ways to the firing of the sense inputs. In my model, the similarities in cognition outside of immediate sense data is simply inserting excerpts and abstracts of that remembered sense data into the scenery in a "play" in room 2c.

The epistemic gap is real, and is EXACTLY the same as the limitations on induction. (I need to read Hume, but I gather he's the one who first discussed it). It is simply that we are small and bounded creatures in an enormously larger universe. Water "could" be XYZ rather than H2O on some distant planet ... it is "metaphysically possible" that there are other elements with other properties elsewhere in the universe ... but that would directly mean that the results of our inductive models of science have failed ... at least until we encounter elements X, Y, and Z and can retrofit them into a more complete model.

Given a materialistic model of consciousness and thus phenomenalism, why would epiphenomenalism be classed as "Dualism-E"?

-- TWZ

At what point does being anthropomorphic stop being projection and start being recognition?

-- TWZ

While being introspective, we see something red. The abstracta filters from scenes 1 through 4a, and because we are being introspective (our present decsision goal set wants information about what we are doing) it shunts the process to 4c which writes a script and sends it back to 2c where we run the play of ourselves seeing something red.

We have real experiences of seeing red, but the experence we have when thinking about our experience is a phantasm. The "red" you "see" when you close your eyes and imagine something "red" is not the real experience of red. It is a place-holder the brain has constructed to stand in for the real experience while pointing to references of the real expience.

Note that this explains the faculty of the brain to adapt to visual shift experiments. The brain isn't rewiring itself it's entire optical machinery, it's just revaluing the reference pointers (possibly with some minor rewiring of some synapses as per the ordinary learning process but not a rewiring of the whole kit-and-kaboodle)..

-- TWZ

My chief problem here is that I think you (and a lot of others) are investing deep magic into the ideas of conscious/consciousness that, in and of themselves as words representing an idea, neither deserves it.

Firstly there are two different phenomena here, the state of being conscious, and the conscious awareness of oneself called consciousness. Edelman also makes this distinction (though I don't remember the jargon - something like first order consciousness versus higher order consciousness) in Wider Than the Sky.

When my cat or my wife is unconscious, they don't respond to ordinary external stimulae. When they are conscious, they respond appropriatedly and "track", i. e. anticipate external stimulae. My internal experience matches that characterization - asleep, I am generally unaware of what's going on around me but awake I have some such awareness and frequently anticipate.

That's ALL conscious is. No magic.

ConsciousNESS or higher conscioiuness, whichever terminology one prefers, doesn't appear to me to be much more complex: an awareness of oneself as an agent in one's environment and awareness of how one''s agency in the environment shapes the decision options beyond one's own action (as opposed to mere contingency).

My model is rather explicitly designed to encompass these ideas of conscious/consciousness.

As per Chalmers, the "mystery" or "hard problem" is not conscious/ness per se but the phenomenal experience while possesed of those states. While I'm not of any work that actually addresses that, and I certainly don't have the understanding of the fine mechanisms, I think the model gives at least a starting point to discusss "what" is happening in a way that cuts away much of the magical incantation crap; if someone comes up with a better model, well hooray!

But we have to start somewhere, and standing around "gee whizzing" ourselves just smells bad and makes our wives bitch about doing our laundry.

I already did, above (though perhaps buried among other posts).

Your GALS as described is almost certainly "conscious" . As described it certainly anticipates the future and responds to its environment and responds to variations from it's anticipation.

As described it may have consciousNESS but may not as well. If for example, it looked at traffic patterns and decided "I" (the GALS system) will not have resources (in itself, not the airports and runways) to calculate all the traffic patterns at 4:00 if I let the planes come as they are vectored now, so I'm going to reroute some traffic to slow things down. Awareness of itself as an agent it it's environment to optimize that environment to its own goals.

This brings up a point of significance ... if the mental resources are always sufficient to deal with the environment, consciousNESS is never required. It is only required if it is necessary to *game* the environment to find one the optimal path for one's own purposes. The human brain machinery may have had the hardware to be capable of consciousness for a long time ... but I strongly suspect that it, as we have it is a VERY recent (last 100k years - maybe less and there may be wide variance in the amount of consciousness as opposed to being conscious among people today) and is enormously spurred by the necessity of gaming other human beings in a highly abstract game (language).

But the more significant point is that GALS may have both consciousness or just be conscious, but it is over a very limited environment. Such "magic" or "mystery" as there is in the human brain (IMO) lies not with consciousness and its mechanisms but with our ability to learn (and implicitly information representation) and exercise that consciousness over a wide scope of phenomena simultaneously at various levels.... and likewise dealing effectively with the explosion of possible permutations of abstractions and representations over that scope.

I doubt any one aspect of cognition is really a "hard" problem. What is "hard" will be understanding how all these sundry mechanisms (whether understood "merely" conceptually, or we actually sort out the grey gunk and understand the electro-chemistry) fit together and maintain balance to keep functioning over a lifetime.

As per my model, we NEVER have the "conscious" experience of seeing red. We see red while we are conscious and react to it, and our consciousness (room 2c) stages the representation of self. The "conscious experience" is the representation of ourselves seeing and reacting to red.

-- TWZ

Yes "conscious" and "consciousness" equate to various levels of awareness about the the outside world and the relationship of oneself to the outside world. Your GALS system is conscious and may have consciousness, but over a limited domain.

The reason you can't "talk" to it is the limit of that awareness. It's "semantic field" is limited to aircraft in the air, airports and their various characteristics, and if it has consciousness at some level, about its own state, present and projected, under various conditions. And within the syntax it understands you can no doubt "talk to it" about these things.

You can't talk to it about the differences between romantic and science fiction because they are not within its semantic awareness - it is not conscious of them. The real "hard problem" (and connectionists are working on it) is representing information, including semantic transfer across a huge range of syntactic representations in a way applicable to the wide range of things about which humans are conscious, and understand their own relationship to (have consciousness of).

Agreed, awareness, thus being conscious and even having consciousness and to some extent even learning over a limited semantic field is often an easy, and not quite as easy but still easy as qualified. Learning over a wide semantic field is NOT an easy problem.

No, no, no NO. I'm not saying I agree phenomenology is (necessarily) a hard problem, I'm pointing out that Chalmers is not defining the "hard problem" as states of consciousness, but as phenomenology.

Yes, yes, YES! That's the point. I'm asserting that there are models of cognition possible (mine previously proffered) under which phenomenology is no longer a dauntingly hard problem, but one which is more easily addressed and can be assaulted with more mundane conceptual tools by which understanding (and perhaps utility in the not too distant future) can be obtained.

And under my model, it is not much of a mystery, profound or otherwise.

Your "conscious experience" of seeing red is actually a representation of yourself reacting to the color red. That is what is really happening is a pointer to contingencies of your self + a pointer to a past saved experience of your seeing red.

Your friend's conscious experience of seeing red is likewise a conjunctive representation which is a pointer to a contingent representation of himself and a pointer to a past saved experience of himself seeing red*.

Now all even though if you swapped the "early" part of your brains for color perception, you would sere green for his red, it doesn't matter because all your contigent experience for red says "red", just as all his contingent experience says "red", even though the "values" for the memories (if the mind had fixed locations in the first place) are different.

The "magic trick" (stage illusion) is that your "conscious experience" of green is EXACTLY the same as your "conscious experience" of seeing red. The ONLY difference is the value of the pointer. Likewise for your friend. That value DOES NOT FUNDAMENTALLY MEAN ANYTHING. So long as the relationships between entities in 2c are correct, the values of the entities do not matter so long as different things are distinctive, i. e. different placeholders represent differently.

-- TWZ

Meta note: Emo, do you mind if I place excerpts from this and the antecedent thread, including your statements in my blog (with pointers back here, of course)? I will want them somewhere I can review in the future.

I'm saying that there are models of information flow under which Chalmers' hard problem is attackable as more amenable problems.

If you have a physical connection to it, and the software to implement the protocols and syntax, presumably you can, but of course only in syntax it understands delivered in formats it "understands" how to create. Once again the problem is not consciousness, or even phenomenology, but only the scope of the semantic knowledge available to it.

In this instance "vestigal" verges on grandiose overstatement, but I agree with the idea in principle. The problem does not scale linearly, particularly when the semantics of interest expand. As per Act 5 of my little opera I think the puppet theater is key to the capability of semantic transfer that is required for general language. As the AI people are discovering the complexity problem is real. It may look like an easy problem, but it becomes hard because scaling is higher than linear.

How do you know it doesn't? You said it does simulations, and can (to posit what I term consciousNESS) can even anticipate its own reactions in such simulations. My conjecture is that "consciousness" is a puppet theater where we do simulations. Because simulations are carried out with pointers and references with only syntactic venue they are by definition not shareable.

Thus my conjecture is one small step up on the scale of abstraction and shows why such experience is not shareable.

And there is thus no reason to suppose that what your GALS system is doing is, in principle, not the same kind of thing happening in your head.

"1" is ultimately the leaf of all the pointers, The actual experience (but not conscious) in your brain consequent to its hardwired connection to peripheral senses

"Something it is like...." Rather telling isn't it? Our syntactic representation of phenomenology as an ontologic entity is telling us rather directly that it semantically refers to a comparison. But there is no external referent, both entities of the comparison must be internal to the thinker's mind for this representation of the "problem" to have arisen.

But now, a comparison of what to what? My conjecture says that the comparison to the real physical - but not conscious - experience of the referent to the conscious representation (a pointer) of the referent. In the normal course of mentation an awareness of the distinction does not arise. But here talking philosophy, we create new abstract entity of this comparison (in computational terms the value of the pointer versus the value pointed to by it) and call it a "quale" - and complain that it has no shareable "meaning".

Even in my model/conjecture this is provisional, but I would posit that the "not in the dark" nature of phenomenal experience is (and only is) the variable nature of the self pointing variable in the simulation. I can think of "red" and think of the shutters on my house, the balls I used to play with in the school yard, my pickup or a candy apple red Harley I've lusted after. I can't think or envision anything else not "in the dark" means except in having the experience of thinking of something with an explicit reference to self.

Furthermore this variable representation of the self in relation to the subject referent is compelling as it hints quite well as to how semantic transfer occurs. We instantiate the reference to ourselves with one set of contingent references conjoined to the subject reference, and subsequently another set of contingent references and note some correlates, and viola, a new abstraction with a primary semantic referent arises.

Of course my model, may be all wet (though it is good enough for me to "understand" myself, and interpolate some of the strange "beliefs" of people around me) but the primary point is that some models of cognition may well render the "hard problem" understandable and addressable by mundane material means.

-- TWZ

Well, a house is a dwelling. A cinder block in a field is a dwelling - for mice and insects. A large pile of blocks and wood in a field is a dwelling - for mice and insects. It doesn't become a house until it's put together in the right way. Likewise no amount of electronics is going to become a conscious entity unless put together the right way. Figuring out how such would be put together is (for me, though I doubt I'll personally ever be in position to do so) what this conversation is about.

Descartes' theater idea has so much appeal because internally it "seems" to match our experience. It is in part, "real" as in there is a part of our minds that "looks like" this. But if you follow my model you will see that there is no recursion, no self conscious entity looking at it. It is just the capability of our hardware to throw together reconstructions of varioius saved real sensoria. And the "audience" is entirely mechanistic with no separate consciousness. It iterates, sometimes many times a second with additional "scenes" but there is no recursion. We can "watch ourselves watching" - much as I am doing in part in writing these posts - but we are essentially time-slicing when doing so. If there were an evolutionary niche for armchair cognitive philosophers our remote descendents might find a second stage built in their brains.

Ok, you have GALS and GALSR run the same simulation. The outputs in terms of semantic references - external phenomena will be exactly the same within the limits of the simulation. You say "red", and I say "red" and within reasonable limits about things on the outside world we agree. But, you cannot save the file of the simulation data runs and expect to get a byte for byte match. Their internal values will be vastly different because they are different hardware with different values for the primary sensory data. We know that all biological hardware and human experience is distinctively different for each instance. I'm saying that my model explains - explicitly - that qualia, DEFINED AS conjunctions containing reference to the VALUES of the internal data are fundamentally not shareable. It is no longer a "mystery" why qualia are not shareable. The very definition, in terms of the data flow I'm expounding, guarantees that they will not be shareable. The "mystery" evaporates. A quale is a square circle.

Of course. We are both capable of generating structural concepts about the abstract idea "what it is like", and likewise for any given external referent. We can "say" that we think they *should* mean something together, but it doesn't mean that we are correct in thinking so. I can say "square circle" and claim that it "should" mean something until the sun explodes. It's a little more subtle because "quale" can be defined as having a "real" meaning in terms of the syntactic manipulations within our minds; but there is likewise a syntactic reference our minds are capable of for "square circle"; there is simply no semantic reference possible.

Of course we can parse both clauses and see that they agree that a tigger belongs to the class "things" and have the property "wonderful". The semantic interpretation of each (english) clause yields and internal syntax in our minds which is identical ... but there is no internal semantic resolution for the subject object of "tigger". Without that resolution we don't know what a tigger is.

Of course again. You can, in room 2c, throw up a pointer reference to yourself, a pointer reference to the abstraction of "experience of redness" and as the correlational references fall through scene 3 and hit 4, decide that "marvelous" should be in there too, throw that abstraction back into the scenery in 2c fall through and decide that it is good and decide to post the above sentences. No mystery so far as I'm concerned.

A soul perhaps?

I used to worry about that, but as time and my life went on, I decided it couldn't be that important.

-- TWZ

Nope. I'm a former VM sysprog. I've had VM systems five deep on a single engine machine. Time slicing. Normal consciousness is two levels deep. Pretty much everybody does it a fair amount of the time. Thinking about the things I write about here, I fairly frequently (though with some care and effort) go three levels deep - and a lot of people have trouble following what I'm saying. Despite being a fairly bright boy I have real trouble going four levels and can rarely make any sense of it. We can talk about it in the abstract (and it's still only second order abstraction no matter how many levels we're talking about), but I'm extremely skeptical of anyone who claims to go four or deeper with any regularity.

Yes if GALS and GALSR use deterministic random-number-generators and start with the same sample data you can write a program that will essentially convert a GALS saved simulation run to a GALSR simulation run and vice versa ... they could share their "qualia" of the simulations. But if they use non-deterministic RNG's in running simulations then such is fundamentally impossible. As per brownian motion and trillions of low-threshold chemical reactions I very strongly suspect that human simulations in 2c are always going to equate to ndRNG simulation runs and thus the "experience of red" half of the quale syntactic structure is always going to be a pointer which is *always* arbitrary outside the system that generated it, and thus fundamentally non-sharable.

-- TWZ

Emo, look at my handle. Not only have I BTDTGTTS, but I sometimes take over the concession stand for a while when the dude there wants a break.

Where exactly to you think I got the architecture for the second level? I have and do sometimes quiet my mind ... and what I "see" are bits and flotsam of stored memories from my senses "like" the caps of waves on a stormy sea. I HAVE meditated specifically on a color (several times): at first all is fine, except that the memories of "things" of that color are always there "polluting it". Then slowly as I suppress that I start to picture a uniform field of the focus color. But a funny thing happens - as I suppress/quiet more and more of the side memories and the unblemished field becomes quieter and smoother. The color actually disappears. Not becomes white, or dark or grey ... though at the early stages it's a bland gray ... but "deep" enough and it's a void with *nothing* there. ( That must be where my soul is supposed to be ohmy.gif )

Thus my conjecture that "conscious experience" is never a primary experience and vice versa. It becomes "conscious", sometimes almost instantly because it is being predicted in 2b, and that pre-simulation can be re-projected in 2c along with re-projections of memories what really happened to isolate the differences of importance (of course they are not real rooms and actually probably involve much of the actual same "hardware" ... the separation in the model is to distinguish the "what" consequences rather than a schematic of the actual hardware "how").

I need to program a key to type "of course". For as elementary experience as "red" your "index" of red is likely going to remain the same (say from about the time you were a year and a half old) or shift only slowly over your life. Likewise your concomitant and correlated memories with it. But it's still just arbitrary values of no meaning outside your particular system.

<of-course-key /> Color by itself, rather than a specific instance, is an abstract concept. It's an abstract class, a purely syntactic construct without its instances. Since we both "do" abstraction and neither of us expects to point to something and say "color", but we can point to something and agree on the specific instances, we agree that we agree on the abstraction. Being abstract, the "quale" of color is "color". We intuitively know that all we can share about it is the definition. We don't bitch about the mystery of not being able to share it any more than we bitch about not being able to share our "quale" of cube-root.

-- TWZ

(new correspondent, mslongjr has kindly given me permission to cite)

Yes! Except for a few "flow" details, you've nailed it. You never *consciously* experience your real physical phenomena - you react to them, which is how you can sometimes catch something you've knocked over long (half a second) before you have conscious awareness of it. But you have memories of them - which you can slice dice and re-edit, replaying them almost instantly for your instant "consciousness". Or, I postulate all of your "consciousness" are parts of these sliced, spliced and re-edited together to represent whatever you are "consciously" thinking about. Do you ever have a conscious thought which is not in some form or another visual, audible, or (more rarely) touch or kinesthia?

Even writing programs, I "see" or "hear" the program tokens an instant before I type them. I never - in that instance - consciously know why I have chosen to do what I'm typing. I can, of course usually recreate "consciously" the reasoning my "subconscious" took ... but when I do so I "hear" "because yKeys is an iterator over the keys of the hash" or "see" the ERD. Consciousness is *ALL* remembered or created (from prior memory, albieit possibly little pieces of many prior memories put together an an apropos way) pseudo-sensoria. But what you see in the theater are glyphs that carry pointers to massive amounts of knowledge about the subject that your subconscious uses - entirely mechanistically, I posit - to do the real thinking the rest of the way "down" before an action (which may just be deciding to remember something) is taken. But you are never *conscious* of that additional knowledge and the processing of it unless the machinery decides that the pseudo-sensory representation of that hidden knowledge might churn additional associational relationships that could be relevant to completing your immediate agenda (E. g. if I "picture" the relation diagram of the database I'm working on in my "theater" I might "see" - in an almost literal sense - a relation between tables in the diagram that I hadn't noted before, which might save me considerable work, or tell me that I need to account for the additional relationship. The information was "in my head" but not in a form that the mechanistic and entirely unconscious machinery could use to tell me how to write the program).

That's *exactly* what it is. Almost everyone who thinks about consciousness stumbles into the Cartesian Theater as the first stage of such. This "model" seems universal - and yet we know, because it implies indefinite recursion, that it cannot be correct. So my model attempts to show why it is a universal phenomenal "reality" and yet does not require recursion. It imputes iteration when the blind machine might deem it useful - and any programmer will tell you that iteration and recursion can both do the same job over a finite range - but iteration (loops) can be broken and can be run mechanistically in the same instance of the current stack-frame (the individual, wholly mechanistic, mind) rather than requiring a re-instance of the current stack-frame (no wholly and itself conscious entity needs be watching the theater). As an additional point, there are a lot of mental disabilities which look an awful lot like program loops "running wild" .

YES! (And perhaps better expressed in a way more understandable to people who aren't programmers.)

-- TWZ

So this mangy dude wanders by, opens a bag and leaves a piece of rabbit shit on my table. What does the rabbit shit mean to me? Did he mean to insult me? Was he complaining that he stepped in some as he crossed the yard? A zen puzzle that the universe has cast just for me? Or that he wants to marry my daughter and is offering his prize bull?

Without the content of his memory, the rabbit turd is an arbitrary object of no meaning. To me your "conscious experience of red" is just another rabbit turd. It doesn't "mean" anything disconnected from the other contents of your mind. A quale is just another rabbit turd.

You misunderstand. A color is a real property (well no it's not, we now know color is an artifact of how we are built, but to go on pretending...) and if we can communicate the separation of the property of a real thing from the object it is a property of we can agree that a color, red, blue, orange, ... whatever, are real "things". But "color" is a collective noun, an abstract entity conceived as a container or surrogate, and often grammatically a pronoun for them. As a slightly more concrete example would be cat/cats. A cat is a specific entity in the real world. If you and I are together we can point to the same object and agree "cat". But "cats" (the category not its polysemous usage for multiple individual cats) as a category is an abstract object.

-- TWZ

What you said in the part I bolded is exactly and specifically correct.

And it's a "Gotcha".

A "quale" is exactly and specifically an intellectual thing. It is about thinking. And about thinking about thinking. It claims that an abstract "thing" which only arises in the context of thinking about thinking "really" exists prior to the conception of that thing.

-- TWZ

No, but thanks for the snark. It prompted me to figure out exactly where meditation practice fits into my model which I've been puzzling over since writing the opera.

I *know* I am not asleep in such. The first time I did this and thinks went void I was startled instantly out of meditative state - my body reacted fearing I had gone blind. Of course everything was fine as I snapped out.

*I* still exist in meditation. *I* am still quite in control - else how could I be choosing what to suppress. But the conscious "I" where we conceive ourselves to be in normal life is what is suppressed. Meditation (at least the zen like forms of it) is the process of suppressing or very tightly controlling the pseudo-sensoria presentations (room 2c), largely short circuiting it and letting the "machinery" take over without distraction.

Thanks.

-- TWZ

ASB, I've put a model on the table. You can falsify it by finding a phenomenon which clearly contradicts it or displace it with a better model. It is admittedly vague in many respects, which is simply an admission that we don't know everything about the details of how the brain works. At least one poster, mslongjr, whether he agrees or not, seems to understand most of what I've said about it. And while vague in many details, it does, at least potentially offer some opportunity for falsifiable experiments.

It "explains" the phenomenal reality of the Cartesian Theater while circumventing the recursive impossibilities in Descartes' original description. (Iteration versus recursion)

It roughly explains how phenomenal conscious experience occurs. (recreation of edited sensory data)

It explains how the mass of "information processing" which we know must be happening in our brains is not available to phenomenal consciousness. (2c is only a small part of the total mechanism).

It explains and delineates several states of awareness and consciousness. (wings and levels, e. g. daydreaming is running 2c hot while derogating the rest of level 2)

It explains why the intentionality of qualia are epistemically non referencable, i. e. meaningless. (reference to a syntactic pointer without semantic reference)

It explains how we have the faculties to react, and react rationally and with intentionality, to events which transpire in a time too short a time to be grasped consciously.(the bulk of our mental capabilities, including understanding of our intentions occurs "below" conscious awareness).

It explains how we react more quickly to changes in familiar circumstances than in novel circumstances. (we generate simulations of what we expect from past experience and react more quickly to variations from the predictions).

It explains why there is phenomenal consciousness. (sensory associations versus syntactic manipulation).

It explains why language is bound to phenomenal consciousness. (rerepresentation of intentionality (meaning) across different senses)

It explains (at a high level) the mechanisms of metaphor and semantic transfer. (reprepresentation of intentionality as different pseudo-sensory representation)

It explains a variety of mental malfunctions, both of the impaired and short term miscues of the otherwise competent. (iterative loops run wild or failure to instantiate a stage, e. g. hysterical blindness)

-- TWZ

We are at impasse.

I am saying that yes, conscious experience is completely explainable as information "sloshing about".

I am saying that in such a system, and even in simulations in your hypothetical GALS system, the "experiences" are the same kinds of experiences. But they are not the same experiences. There is no way, even in principle to compare the actual phenomenal experiences of two different individuals, and it is this impossibility which makes "qualia" a chimera. Given the individuality of human brains it is extremely doubtful that any two human beings, anywhere, for all time, even identical twins raised together have (even presuming we had the tools to test it) the same experience of anything. In fact, given what Edelman and others say of how the human brain is built and functions, it is highly doubtful that any single human being has the same experience twice.

It would be a tour-de-force bordering on farcical, but we probably could today build a system as you postulate, and if we limited it's discursive domain to color "experience" and simple metaphors, have it pass a limited Turing Test in discussing such.

As I understand what you've written, you are saying, you would not credit such a system as "understanding" color experience no matter how often it fooled its human judges.

This is particularly amusing as you assert that I "seem to be missing something", and yet I misdoubt that you would question my humanity.

I recognise that you will disagree, and duly credit it, but from where I sit, your statements seem to be saying "there must be magic in there somewhere".

It states quite clearly that the experience you enjoy - i. e. have consciousness of - is a simulation built from memories of your real time (and not directly available to consciousness) sensory experiences along with a contingent (often just a pseudo-literal point-of-view) reference to yourself.

Our instinctive reactions bypass the need to invoke the conscious experience because that is too slow. Our knee-jerk reactions are part of the "easy problem". They function by recognising patterns, detecting emergencies and running an appropriate response algorithms. They can also be trained by Natural Selection, The Baldwin Effect or by Conscious effort. Very neat, very clever but "easy".

My model gives a little bit of "how" and more detail on "what" is happening in such circumstance Your statements are simply an acknowledgement of the observed phenomena about our reactions exist and speculation on how they came to be in evolutionary terms without explanation of how it occurs in present time.

The model is a denial that the hard problem is hard. It is an explicit attempt to reduce the hard problem to more workable pieces.

Not even close. The model is a rough picture of data flow in human cognition to account for the phenomenal experience of consciousness.

Another area of disagreement between us the "hardness" of "information processing" across layers of representation, reference, "meaning", learning, fixation all through the pure syntactic processing in levels 3 and 4. I agree that over limited domains of information it "looks" (relatively) easy. But the people trying to implement it over a wide domain - human experience - say that it is not so - though they do seem to be making progress.

And we haven't touched the complexities of human emotion. I agree with my pal, charlie-d, that emotional cues are likely very fundamental drivers throughout the entire system - I simply think they and their motivational impetus is coopted and transferred throughout the "information processing" pieces as relevant to their informational structure.

Which is part of why I think "putting it all together" is another, as yet unacknowledged, potential "hard problem". We may be able to build a useful, interesting (facinating) AI beforehand, but we won't have one that thinks "like" a human until we have a good low level understanding of how our basic motivations interact structurally with the information flow.

The premise of this model is that such is fundamentally impossible. Consciousness *is* part of and arises (solely) from information processing in the brain.

A concrete contrary hypothesis, other than gee-whizzing, is what is missing here.

It's something like GALS' radar "seeing" a thunderstorm twenty clicks southwest of JFK.

-- TWZ

Roughly, yes, this is our impasse. qualifying "information dense" as actively capable of running simulations on input data with the conception of self as an agent which will affect the sensory data received.

For minimal circuitry a "conscious" circuit could be built with a photo receptor, and perhaps three nand gates, outputting current state, and varies from anticipated output and an "anticipated" state input. A consciouNESS circuit would have an input of current (simulated) state, desired state, and additional INPUTS concieved as controlling a light in proximity to the photo sensor of turn light on, turn light off and do nothing, and a "LEARN" input and a single output "right/wrong" or yes/no. The consciousness circuit would not be hardwired directly via nands to the outputs but woutd be 32 bistable flip-flops (64 NAND gates) plus a handful of others to direct the "address" correctly.

Below this is a logic heirarchy (trivial, easily implemented on a lowly 4004) connected to both of these which, if the light and photo sensor were actually connected (or likewise trivially done in simulation) which would do the action of turnin.g on or off the lamp and generating the "LEARN" inputs to the "consciousNESS" circuit.

Adding some trivial programming, you can "talk to it" about what it "would" do to achieve various ends in various conditions, and about whether it is actually seeing what it anticipates.

As I see it this entity has exactly the same KIND of consciousness we do, and in the consciousNESS circuit it has the same KIND of experience we do (simulates what we will see and what we could do and decides whether the result matches desire)

The difference is ONLY the huge range of sensory (including emotional) information from outside that part of the cognitive system and the VAST breadth and depth of semantic scope derived therefrom in our "circuitry" used in the simulations. This is why I keep stressing the "hard problem" of saving and using that information consistently which you deny is "hard" and Chalmers glibly classes as "easy" is so important.

Such "magic" as there is (IMO, and again I understand you disagree, thus we are at impasse) is in the complexity and depth of the information processing piece itself, not with either consciousness or experience.

No but I will argue that having a model - which I cheerfully admit could be flawed in many ways - is more USEFUL than gee-whizzing. Fortunately AI people, neurologists, and philosophers are using various materialistic models (again, "my" model is what I've read of others - mostly Dennett - have said and putting the pieces in little boxes and drawing arrows between the boxes) and doing experiments to empirically sort this stuff out, and build better models going forward.

Magic. That's all.

-- TWZ

How exactly can a description of human like characteristics in a non-human implementation not be anthropomorhic? The word in such context just means recognition of similarities to human behavior.

Those words are not per-se anthropomorphic but heuristic descriptives. The circuit is "conscious" because every primitive heuristic (I see, no doubt you and philosophers can have a ball arguing) of consciousness qua consciousness has a correlate in that circuit.

The circuit can't describe those heuristics, of course, because it has only one bit of awareness and understanding heuristics requires a wealth of information and knowledge to build it. It's only heuristic information is whether the light is on or off and what it can do about it (again, described from outside by entities with far greater understanding of the environment where the circuit encounters it's little problems).

And you don't "see" what your brain is doing in terms of sloshing chemicals about and firing synapses. You "see" an outside problem you interpret from your senses and conceive of what you can do to change those circumstances. The circuit "sees" only the light and has the heuristic knowledge of what it can do to change whether or not is "sees" light.

Once again the ONLY difference I see is the wealth of information available from our senses and the store of that and the depth of information we abstract from it.

I assert that we are (except for signal noise due to brownian motion and various hardware errors and software glitches) every bit as deterministic as that circuit.

I know exactly what NAND gates are. I used to build them from transistors and resistors way back when I still had to dodge the dinosaurs and trivial TTL logic gate DIPs were several dollars at Radio Shack, but I had a store of transistors and resistors "rescued" from discarded radios and such. NOR's are easier ... I had to wire three NORs together to get a NAND.

We (here) haven't begun to discuss properties of heuristic interest in that circuit. How competent is it? How adaptive and how complete is its learning program? Can we get a higher level of abstraction by adding knowledge about the do-nothing versus switch-light-on when it is already on? Etc. There is a world of correlation to known "problems" and issues regarding human minds which can be discussed by incrementally adding features and capabilities to such a simple circuit.

Because you declare, a-priori that phenomenal experience is:

That's exactly what evolution did.

-- TWZ

And I have said before that the missing information in that metaphor that 0!=1 is the hardwired sensory primitives.

Where we differ is that I see "seeing red" as pure information, absent any conscious experience. It does not become a conscious experience until and unless cross correlated with other memories one has of oneself.

I see absolutely no reason to presume that sensory reciept of informationm say red impinging on our retina - by itself - carries any more conscious correlate for us than a contact with its cilia does for a hydra. Or the conscious subcircuit of my model one bit consciousness circuit.

The difference is what we do with that information, and we do it with a massively complex deeply brain which stores and builds massive and deeply organized models of itself and its prior experience.

And I repeat again that my model and subsequent discussion is an assertion that Chalmers' hard problem can be rendered as more amenable problems.

You've agreed that such is a reasonable goal and should be possible, but have failed to provide an alternate means of doing so. Everytime I start pushing the specifics you back away and complain that we need to separate out the easy from the hard problem. If as the assertion hypothesis - to which you agreed is possible in principle - is correct then there is no hard problem at all, we only need find the organization of easy problems that will solve it. I've offerered one - you disagree, and that's fine, but you haven't offered or pointed to an alternative. You've only fallen back to the liturgy that "experience" is some magical hard problem.

No again. There has hitherto never been a mapping of human conscious heuristic terms to engineering jargon. THAT is the one and only (IMO) reason that you cannot describe "the hard problem" in engineering terms. This thread is an explict tentative attempt to do so.

And I've done EXACTLY that. I've given you a circuit that I assert IS exactly that.

Rather than bitch that you don't see it, tell me what you see that that circuit doesn't.

What can you do with one - and only one - bit of information and heuristics for dealing with it - but only that - that that circuit cannot do? What can you do with "the experience of seeing red" without the other information in your brain?

As an extra credit problem, consider the possiblities of say a room full of such circuits configured so that each can see a few of the others' lights and some variability in the learning programs driving the 4004.

-- TWZ

HSP? Not sure what that is, but yes, as I see it what the brain does, other than making our heads large enough to fit our hats, is information processing (with our biologic imperatives, which we "feel" as emotional responses, being ROM, metaphorically).

Yes. Depending on how deeply color perception is hardwired in (pretty deeply, I suspect) it may take only minutes ... or it could take days. I suspect more subtle perceptions are very likely in the months. But the ability to recognise it as a conscious experience will always be learned and not hardwired. Dennett, in CE and possibly some in TMI had some discussion of hysterical blindness. It's a classic structural failure in terms of the model - the raw perception is still there, but the brain refuses/can't build the correllational models. In some instances - the "blind" person can move around without often bumping things, thus the anticipatory model is still there.

Not at all, but it is a bit of insufficient communication on my part.

The conscious subcircuit is not by itself conscious. The entire mechanism, including the 4004 and its programming is conscious because of the action of that circuit (which, of course, could be easily emulated - except for the photosensor - in the 4004 itself, but is here limned as a separate circuit for purposes of explanation). And the whole circuit is only conscious when it is determining the anticipatory state and paying attention to the variance output.

As limned, in terms of pure receipt of a single bit of information, absent an anticipatory response (generating the anticipatory signal) the input is pure information - high or low signal from the sensor which the programming of the 4004 may or not be doing anything about.

Quite the contrary. I only deny that it is magic or its presently politically correct synonym, "hard".

My model - and the one bit consciousness circuit - implicitly defines consciousness as the process of correlation of mapped sensory input to past experience and contingencies for action to change that sensory input. The one and only reason you can identify red as a conscious experience is because you have the memory of NOT seeing red and know that you have some ability to affect that condition one way or the other. I also point out that these are not fixed states but process semantics. The circut's conscious and consciousness status is embodied in the capability of generating the inputs and then later reading the result over a variety of conditions not a single fixed state in time.

I certainly don't deny conscious experience - I seem to have a lot of it most of every day. What I deny, pending contrary evidence, is that it is anything more than what lies in the scope of the above description cum functional definition.

-- TWZ

See Acts 4 and 5 of my opera. Evolution discovered that animals that had the capability not just to anticipate the future, including contingent paths of its own actions, but to anticipate the outside world's reactions to that animal's decisions and account for it in planning survived better.

But such models require a placeholder for ourselves as actors. That placeholder - puppet of ourselves - in these models is "free" ... it can move about in a model world in all kinds of ways and do anything at any time, but in sane, USEFUL models its actions become more and more constrained as the time and events of the puppet in the model and real time and events converge.

The essential illusion of consciousness is that we are the puppet in the model world rather than the reality of the fully determined meat robots of the instant moment in the real world. But this power to model the future and take action to find ways to make the real world in accordance with our wishes rather than simply finding what near things are nearest available to our wishes has given us dominion.

We are simultaneously and in genuine little t truth with no paradox at all unconscious automatons acting deterministically based upon our prior inputs and free willed fully conscious agents changing the world in accord to our wishes.

Consciousness is not knowledge of the world as it is and what we are doing, but knowledge of what the world may be and what we can do, not in this instant, but in the future from a split second to years yet to come.

-- TWZ

I realize that you don't "buy" my model, but I am unsure whether that is simply a visceral reaction to your belief in the ineffable nature of "experience" or a failure to understand how it works. Once you understand how it works, it is easier to let go of the "magic" property of experience and see it for what (I believe) it is: the disparity between the real mechanistic self and the projected illusion of self.

So let's do a walk through ...

Now EVERYTHING in this is purely mechanistic. We don't know the details of the mechanisms that generate visualizations yet, nor the comparators and syntactic logic that generate programs for the automaton us to follow ... but as per Chalmers you seem to consider this an "easy" problem. It's just taking memories and recombining them in various ways. This machinery obviously conflates current desires and memories of past fulfilment of those desires with contemporary sensory data (very short term memories of them actually, as per my model) about the outside world and looks for correspondences that can be turned into internal programs to drive the meat robot around in the real world to satiate those desires. Now as it's just "information processing" you tend to dismiss it as an "easy" problem ... I disagree.

So ...

sense: Hunger.
sense: Stream nearby.
visualize: Fish in stream.
visualize: Self eating fish.
recognize: Good.
program self: Wade into stream and catch fish by hand.
execute: This involves carrying out a few thousand iterations
	of runs of lesser programs just like this one, but 
	"this" run had the knowledge of self to know that it 
	could generate and execute those subprograms, so it 
	wasn't concerned with generating them at this level. 
	It simply time-slices those lesser programs into 
	"this" run, and our story continues ...
result: FAIL
sense: HUNGER!
visualize: Eating last meal (rabbit);
visualize: Hunting last meal
visualize: Rabbit impaled on spear.
transpose visualization: FISH impaled on spear
visualize: Self eating fish.
recognize: GOOD!
decide: Hunt rabbit or fish (just compare time and effort 
	and success likelihood of the options)
program self: Fetch spear and return to stream to "hunt" fish.
result: ...

It just might work more often than fishing by hand and in some cases more than hunting rabbit.

Our hero, a robot "mindlessly": carrying out its programs, has just expanded his "free will" in thinking about options for "hunting" in the future, simply by a simple transposition in mental imagery. More he has vastly improved his chances for survival if he badly sprains an ankle; he cannot run down game nor stalk quietly across the grasslands, but if he can hobble over to a stream he stands a chance of feeding himself and others that day.

-- TWZ

You write your program or engineer your system to achieve that desire and slake that hunger. These again are heuristic descriptives of the programs our brains and bodies run. My computer has a "desire" to print out my taxes when I click the print button. Whoops, now it "wants" a new ink cartridge in the printer.

When our bodies sense that our energy reserves are low the brain invokes a program that causes us to seek energy replenishment. The brain represents that sensory data to us in 2c along with a host of contingent representations. Being fairly common and fairly consistent we quickly generate the internal glyph of hunger to characterize this condition. When we learn language, and recognize that a host of other creatures exhibit similar behaviors that seem consonant with our own, we adopt the transposed sensory representation of "hunger" to represent the characteristic of the state in the abstract problem (requiring heuristic descriptors) of seeking replenishment and map it to our own internal representations of that state. Desire is a wider abstract class including hunger.

So the program doesn't really "desire" to print my taxes subsequent to clicking "print", that's just what it does. But within the heuristics of the problem of printing them out it "wants" the ink cartridge, but at the next level down, it doesn't "want" that eaither, it just stops and pops up a message. But going back up, it is perfectly legitimate to say that it "wants" a new ink cartridge, and it "wants" a click on what to do next. All heuristic descriptive which apply across levels of representation of a problem.

-- TWZ

You ask me to show you how a human condition word relates to engineering a system. The moment I do so, you scream "anthropomorphic". How could such an answer not be "anthropomorphic"?

If you don't want the answer don't ask the question.

Words like "want" and "desire" are not definable within the system that "experiences" them - they are primitives to that system. If I never learned a word, I would still experience hunger. I don't speak a word of Chinese but no Chinee would doubt that I experience hunger, thirst etc. But such words ARE definable at the next level of abstraction in terms of the intentionality of the system - heuristics. In writing programs my in-code notes will say things like "now we want ..." and the utility of using such words is unquestionable; the heuristic "try" is now part of the semantic structure of almost all programming languages developed in the last twenty years.

Our "system" is engineered by evolution with the "intention" (and even that is a heuristic of thinking about the consequences of evolution in the abstract) of preserving the gene. The "intention" of a system you design is up to you.

-- TWZ

I did exactly that, my one-bit consciousness circuit. You disagree that it is conscious and has consciousness; that's OK, you're allowed to be wrong. biggrin.gif

The problem is that you assert that thinking of only one thing is the same as having consciousness of only one bit. I assert that it is not so. Our brains run continuously correlating past and present information from our senses. We CANNOT "think" of only one bit of experience.

Let's ask the more general question. I and a friend come over for dinner and drinks, we get into dorm-room all night bs mode and spend the time yakking various things from consciousness, to women, to baseball, and dirty jokes. You would agree he is human, no?

Then come the dawn, I wave a star trek gadget and the back of his head opens up and you see a mass of wires and circuit boards.

Did he still "experience" the same things you thought he had a moment before? Obviously his experiences were not the same as yours, but neither were mine. Were his experiences the same kind of things as our experiences?

-- TWZ

Correct, but that does not mean that the question has not been addressed. You asked for how to characterize desire in system terms. I gave you that. You claim I'm being anthropomorphic; true but that is exactly what you asked for - an engineered system is like a human system.-

The exemplars of "desire" in heuristic terms at least put us on the road to being able to define "desire" as process semantics - and approaching mathematical terminology: For any process P instantiatated in system S which interacts with it's environment over an extended domain (e. g. time) the state transitions, s, in S accomplished by P with an adequacy of resources may be said to be the "purpose", or "intent" of P with regard to S, and S may be said to "desire" the state transitions, s, accomplished by P. If an environmental insufficiency of element e or anticipation thereof results in cessation or suboptimal modification of state transitions, s, S may be said to "want" or "desire" resource e.

Admittedly this does not cover the range of "desire" in humans - it can't well characterize the desire for baseball tickets for the coming weekend - but it does seem to cover primitive desires in humans, and in an engineered system the state transitions, and environmental requirements can be easily represented as characterized vectors of quantities.

Can't you see that you are so tightly coupled to the illusion of yourself that you can't uncouple the heuristics of the puppet abstraction of yourself from the primitive sensoria of your real self to such an extent that you cannot realize that they are two very distinct and separate things?

But they are separate and thus the abstract heuristics of "desire" can be uncoupled and defined separately from biota without it being a tautology.

-- TWZ

Where in my car is the mechanism of transport? In my TV for entertainment? Point to flying in the blueprints for a plane.

Heuristics are about what the system does, and the mechanisms that do it are all of the components that enable the system to do that.

And we were "designed" by evolution with the "objective function" (as seem from a heuristic view) to preserve our genes.

And the system for "knowing what it is for" - the ability to develop heuristic representations of our primitive sensoria and mental states - is part of the mechanism evolution designed into us to preserve our genes.

And that's all we do ... astonishingly complex algorithms, to be sure, which themselves generate new algorithms on the fly, but vice "evolutionary programming" et al, we're learning to do that now.

And as stated before per my model, you never have the conscious experience of seeing red. You have a real, but unconscious, experience of seeing red, and in your internal representation of yourself you have a pointer to that memory. Since you now have a representation of yourself you can develop heuristic words for all the feely things inside. Since sensory primitives and mental states cannot be shared, meaningful language REQUIRES that we have this puppet self in a model world about which we can generate words which are fundamentally heuristic ... "experience", for example.

We ARE machines. That has been my thesis since day one.

I have been posting what I consider strong (I realize you will disagree) plausibility arguments in support of that thesis. In particular you attempted to use "feely" words like hunger and desire to counter. In response I gave a preliminary formal system semantics definition of desire which is entirely "new" as far as definitions go ... such "new" information precludes the possibility of the argument being tautological.

It will be a long time yet before we have the complete mechanisms of human mind and consciousness reduced to mechanism (even conceptually, as I have been doing here), but I do not see any fundamental barriers; it is simply a case of unravelling incredible complexity. There is a lot that we need to - and will - learn in unravelling that complexity, quite possibly including new means of information storage, representation, transformation and association, but I see nothing that requires any fundamental new principles. This is where you and I disagree and are at impasse.

-- TWZ

Will you please read the model before repeating this canard yet again?

The puppet is a hollow shell with nothing inside, the flickering lights on a TV is a better model. It has no consciousness and moves only as directed by the unconscious machinery of the rest of your mind.

The rest of your mind is also completely mechanistic and lacking any consciousness, but it does have vastly complex machinery for "syntactic manipulation" ... including syntax for generating new moves for the puppet and scenery for the stage based upon past real sensory experience and derivatives from other puppet theater it has previously staged.

Consciousness IS - and only is - the stage and the puppet dancing thereon. (it exists to recreate past sensory experience to generate new associations in syntactica from information received after the original experience, or - lately, in evolutionary terms - to try new configurations of imagined reality as to test them for utility). The stage, puppet and the rest of mechanistic self repeatedly writing and running "plays", modifying them based upon newly generated associations and evaluations and running them again, and then yet again, is the process of consciousness.

Again the puppet has no consciousness - it is empty. The rest of the mind has no consciousness. The unconscious mind watching the (unconscious, empty) puppet dance or seeing the puppet watch the (recreated or imagined) outside world IS consciousness.

The rest of the mind knows - syntactically and unconsciously - that it is contained within the creature REPRESENTED by the puppet.

The rest of the mind knows - syntactically and unconsciously - that the theater is wholly a creation by itself.

The ONE AND ONLY representation of self available to the mechanistic, syntactic mind is the puppet in the theater.

Therefore the syntactic process - incorrectly - ascribes the theater to taking place inside the puppet's head. .. which gives rise to the impossible recursion of Descartes' original theater model that he could break only by postulating soul stuff. But the recursion itself, of course, is an illusion, a mere misattribution, but for the huge majority of human thought throughout evolutionary times, a harmless one. It is only here and now, trying to understand ourselves in purely mechanistic terms, that the distinction matters.

I agree with that completely.

As yet, science has no accepted consensus on the mechanisms of mind and consciousness.

You, and other "mysterians" basically assert that we need to sit around on our thumbs waiting for some breaktrhough to discover a new principle. (Ok, that was invective hyperbole ... I know quite well you support ongoing research into the mind, but you are looking for a breakthrough principle - one which I don't think is there to be found).

But other philosophers - Dennett in particular (interpreted through my rose colored glasses of personal introspection and the watercolors of my vocation) have models on the table that I think are robust and conceptually nearly complete. There is a lot to be done in terms of verification, and huge amounts of work to be done (and vast amounts to be learned) in unravelling the complexity and details of how the pieces fit together, but no breakthrough is needed to "understand" it.

As said, science hasn't a consensus as yet on which view is correct.

And I have given you my one-bit-consciousness circuit. You disagree that it is what I purport it to be.

You think there is something mysterious in need of explanation in "the experience of red".

I think your "conscious experience of red" is just you (the real you) looking at a puppet of yourself looking at something red and the "mystery" you are looking for will be found only when you attack your car with wrenches and a hacksaw and actually find "transportation" inside of it.

As said before, we are at impasse, and as yet science hasn't a consensus on which of us is correct.

-- TWZ

You pull this wire and the arm moves. You pull that wire and the mouth moves. You jiggle the cross piece and move it above the stage and the puppet walks. The scripting is sufficently detailed to say which wire to move and how to move the crosspiece; executing it is just mechanics.

It's better to remember what the stage is, a simulacrum in which excerpts and edits of real sensory experience are reprojected. For visuals a 2D space (with occlusion, thus "2 1/2D") - and at startlingly low fidelity, trusting my introspection. And for sound and others simply splicing them in in proper sequence syncing the track.

Remember my little play about our hunter hero? "The rabbit on the spear is at 247, 322 - 298,403", "Cut that region" , "rescale fish image to that size", "paste at that location" ...

Exactly the same part of your brain that - unconsciously - takes data from your senses and turns it into structured, syntactically manipulable information.

thus:

No, but if it had a camera where it could take video of the outside world, and a computer attached that could take pieces of that, produce its own "shows", show them on the TV, watch them there (ignoring scan line interference and such) with a camera and compare them and subsequently generated internal information to both outside world and its own previous productions, then THAT system would be-conscious/have-consciousness, and the TV would be the consciousness (or at least the instrument of consciousness) of that system.

class Theater{
...
public Theater(Sensoria s, Puppet p){ ...}
}

class Puppet{
...
private MyMind self;
public Puppet(MyMind s){
self = s;
....
}
}
class MyMind{
public Sensoria sensoria;
public Theater theater;
public Puppet puppet;
private SyntacticStructure syntactica;
...
MyMind(){
sensoria = Sensoria.getInstance();
puppet = new Puppet(this);
theater = new Theater(sensoria, puppet);
...
}
}

Another commenter, Sabunim4dan, weighs in with his opinion that discussions on conciousness are premature before the biology of the brain is fully understood ...

Nice analogy, but to push it just a little bit farther, we DO have (low res) satellite pictures of the territory. But we know the vegetation and geography are vastly different that what we have encountered before, so it's extremely hard to make sense of it. Nonetheless, some cartographers (Dennett) have interpreted those images in a way which seems to give a consistent map of the large scale topography. It can be used as a guide to help plan explorers and surveyors (until/unless proven irretrievably wrong, of course). Agreed a lot of people need to spend a lot of time hacking through the jungle before we know just how good that "map" is, and there will undoubtedly be places where it is miscalibrated due to local variance of the conditions, and won't be corrected there (and may in a few instances prove deleterious) until people with laser sights and theodolites are on the spot to correct it.

...

-- TWZ

It will hardly be a surprise that I agree with Dennett. I *know* the mind lies to itself, up, down and sideways all the time ... especially with words. The qualia debate classic in this regard. The power of second level abstraction gives us enormous power and potency - it lets us reduce abstractions to concrete pieces - words and sentences for our thoughts. But our penchant for creating abstractions where they don't apply. It's not a problem regarding things outside ourselves where the abstractions have distinct semantic meaning that can be tested, but it's a nightmarish bugaboo inside our heads and "qualia" is just one of the insects infesting it.

Let me demonstrate with a rather concrete example. I can posit a "quale" for riding a train. It "means" something: I can describe what it's like to ride a train, and that clause is a comparison I can make. I can say it's like riding a car except it's roomier, ... there are more people ..., you can use the restroom... etc. You can talk about "what it's like" to ride the train because the abstraction of "ride a train" has many real varied and external associations. But you cannot talk about "red" that way because the word itself ultimately denotes an abstracton which exists only internally and is not shareable. But I can say - and it has meaning "what is the red of my pickup like" - and answer, "it's like the red of the shutters on my house but lighter, like the red from the pen in front of me but darker". Your (and Marvin's) assertion that qualia are a lingusitic mistake is essentially correct, but based upon the fact that "red" in the abstract, apart from a specific experience of the outside world is a purely synthetic and internal property; by defining it as a "pure" principle it is removed from the possibility of a comparison, thus there can be no "like".

Abstractions about abstractions are what we do - language is utterly dependent upon them - and thus we "believe" them "real". Well they''re useful enough that "real" can be termed to mean something when deconstruction of the abstraction sooner or later points to something outside ourselves. But it doesn't mean that such abstractions are necessarily "real" in all cases, particularly when that abstraction is DEFINED as something which cannot be reduced. "Quale" is a purely theoretic abstraction, and when it's object is DEFINED as something which can never be pointed to, its "reality" is simply the mind lying to itself.

Amen.

I haven't seen or read McGinn's mysterian theory and don't know upon what it's based. But casual contemplation I do have a couple thoughts.

The first is that phenomenal mysterianism is .... defensible. Even if at some point in the (probably distant, if our civilization lives that long) future we can read thoughts and experiences and reproduce them at will, even modified as we see fit, those who experience them will be able to assert that we still don't know from one mind and instant that it is the same for everyone. But such an argument is essentially as meaningless as asserting that we don't understand hydrostatics because we can't predict the exact location and orientation of every molecule of water in a pipe.

The closure argument, I haven't read enough detail to have a lot of thoughts upon, but aside from being equally defeatist as the "god did it" arguments, I have a rough outline of a counter: whatever, and however our minds are constructed, they were done so by evolution. If we can understand what evolution is doing to a satisfactory level of understanding - and we can, though it is often vexing to tease out the details - then we must perforce be able to understand our minds to a similar level of satisfaction. We simply need to figure out what evolution did - or could have done - to produce our minds.

There is no human undertaking where our understanding of logic and information is better demonstrated. I've worked on systems with well over a million lines of code (modularized into much smaller pieces, of course) the depth of logical catenation and contingent "thought" - that regularly works, often millions of times a day for any one such system - in such, even in "mundane" business applications, far, far exceeds the conditionals and contingencies of ANY other human endeavor. Nor is there any other human practice or technology which can give us a test to tell us when we're on the right track in developing models of the human mind. Even Edelman has his Darwin robots (irony). In short we don't/won't understand our minds until and unless we can model it in computers with a high degree of verisimilitude.

The problems are that, as of now our computers have several orders of magnitude less internal bandwidth than the human brain .... but that is likely to be corrected in the not too distant future, at least for top end research machines. But the larger problem is, as per my prior posts, we do NOT yet know how to manage the scale, scope and complexity of such information over a broad range of knowledge, and particularly across levels of abstraction - and more importantly to build new "information" from primary inputs over such scope, scale and complexity, while properly managing what to keep and what to throw away (thousands/millions/?billions+? of times more potentially than what needs be kept). And likewise, but related as I've suggested I expect possible problems of putting it all together in a consistent, "sane" package even when we solve the complexity problems. But I do think the overall "flow" will be similar to Dennett's (or my, pretty much the same but laid out more like a computer program RD) model - but of course we won't really know until we have all the pieces to put together.

Agreed, a model that's correct MUST be consistent with what the hardware can do and ultimately does do.

-- TWZ

Another commentator ...
----

Hmmm. The questions of consciousness and self are literally life and death for thousands of brain damaged people throughout the world, to say nothing of the ethics of our treatment of other species.

I think this thought is generally indefensible, save in cases where the subject is completely known and exhausted ... and even then it should be revisited from time to time to see if new knowledge changes one's past interpretations.

And finally, you're flying in the face of some pretty venerable obverse philosophical opinion in this particuluar topic:

Gnothi Seauton

... from the inscription over the Oracle of Delphi ...

and that exhortation is ultimately what this thread is about.

-- TWZ

Sabunim4Dan again

------

Remember on another thread I offered the tentative definition of information as predictive correlates between disparate physical systems. So we have the disparate systems of the human animal and his envrionment; there seems to be a lot of prediction going on there on the part of the human, ergo he contains a lot of information - does a lot of information processing about his environmnent. And of course we can now do a conceptual separation between the sensory systems and motor systems of that animal ... and what lies between is the brain.

If what causes the predictive aspects of human behaviour, which is pretty much all of it above newtonian laws of motion and respiratory biophysics, is not "information processing" then it is PFM.

Your real question here seems to be whether a machine which could mimic human behavior with total verisimilitude would otherwise have phenomenal consciousness. Now as per my model, yes, simply to be able to converse in colloquial language with appropriate (simulated, if you please) emotive reaction and a full grasp of metaphor and idiom, including ad libbed variants which are common in casual conversation, yes it needs to do something similar, I think, to what we do in our consciousness. Now I think, as per my model, that the mechanistic consciousness that would be required for such is possible and IS the phenomenal consciousness.

But if you think it is possible but question the phenomenality of it otherwise, you are invited to read the discussions (I think Myron gave a link way up in the thread) to zombies.

Sorry but you're epistemologically at sea here.

As per zombies, in current world the Turing Test is the one and only observer independent test for consciousness that we have. You seem to pass the "Turing Test" of discussion on a web bbs, but for all I know you are a zombie.

Of course I think we are all "zombies" and that the phenomenal conscioiusness is nothing but edited and reconsidered sensory data which we process again as if it were "real" sensory data. Whether my model or some other materialist "information processing" model is correct, your aliens, if they have observer independent tests that can identify consciousness in the human brain (other than by assumption from an extended conversation) then they know what that structure of information flow that creates it is and can test for it in an observer independent manner. Thus if we successfully implement a conscious machine, your aliens would be able to examine the data flows and structures and determine that it too has that.

Re your prior message. What "binding problem" do you find difficult? The epistemological version is trivial for a programmer; everything inside the conceptual boundary of the senses is "syntactical" in Searle's sense. But the bottom most syntax references the raw sensory data from each sense, and all the other relationships build from there; so long as the pointers and references are maintained with integrity, there is no "binding" problem - whatever we think syntactically can be traced ultimately to something on the outside to put it together (in neural net logic this is considerably more difficult to find and identify than in discrete logic, but as I've said before dl can implement nn logic and nn can implement dl logic; it is not a conceptual hurdle). The "unity" problem we've discussed before.

-- TWZ

To recap here. Searle's "unitary consciousness" is largely a figment of his imagination. When I "watch" my awareness, I see that it shifts, often several times a second. When I do meditation where I've purposely sought a place with minimal sensory distraction and avoid focused thought, I am aware of what must be literally hundreds of fragmentary thoughts churning below the surface in any moment, each "trying" to "grab" the focus of my mind. When I am focused on a particular problem, my awareness still shifts frequently; perhaps not as often, but the primary difference is that it keeps returning to my present problem.

The brain multitasks, much like a computer, (remember we build our computers in aid of our minds to be like them as best we can - brain multitasking came first!) and the brain has mechanisms to keep track of what instant of awareness "binds" to which problem.

Third the "Cartesian Theater" is (per my model) real .... but it is iterative and not recursive, and the "audience" is not a complete self or surrogate self but the same (unconscious) sensory evaluation mechanisms that survey our real senses. But the brain knows (syntactically, references and pointers, however that is accomplished in goo) that these are scripted plays and the associations arise from the pseudo sensory experience are bound to the particular problem which caused the script to be written. It scripts these in temporally significant order and syntactically binds the "new" information back in sequence. Our puppet self - the only part of us that can describe these experiences - only "sees" unity because what it sees (pointers back into unconscious memory which can be rendered conscious by running another "draft" of the script) is scripted to be part of solving particular problems.

Finally, it all happens in our heads with pointers back to the mass of history from which it arose. It is every bit as informationally re-entrant as it is biologically - hardly a surprise as being ever so mechanistically inclined, it is that biological reentrancy which entails the informational reentrancy. Where you and I differ - perhaps - is that I deem the biological reentrancy does not "give rise to" the informational reentrancy, but that it is the informational reentrancy.

-- TWZ

Neither Dennett nor I deny that there is a component of human cognition colloquially known as consciousness. The process, as part of the other processes in the brain, exists, is real, and has real consequences to the observable behavior of human beings. It is complex and difficult to think about and difficult to disentangle from the complexity of other processes in the brain. But it is in no way magical, nor is it mysterious - I have high confidence that it can be explained as "information processing" - in fact I think in broad outline Dennett has explained it as such.

There is one rather ironical case where the (epistemic) binding problem is a genuine and real problem. And that's the problem.

The qualia of sensory primitives are a philosophical bugbear PRECISELY because they are not bindable. Just as "97FYC2" may mean a trailer in a lot somewhere when displayed on my computer screen, it means that precisely because there is an unbroken chain of abstractions leading to a real experience of someone physically reading (and ultimately writing) that string of tokens on a real object in the physical world. All our mental constructs ultimately trace to similar real sensory experiences. They are ultimately bound (possibly with some internal wiring which makes some correlations inevitable) to our senses which generally point back to the real world. But a quale is DEFINED as an abstraction without referent. It is DEFINED to be not bindable. See the problem? You can have a real, really bound, "qualia" of looking at the red of my pickup truck, or other particular red object but not a real "quale" of the abstraction of those common correlates when the quale supposedly exists independent of the real sensory experience. (And the answer is ... (drum roll) ... it doesn't).

Eeek, The mysterian prince may eat me. I wait in abject terror. (grin)

-- TWZ

This IS the zombie argument. And addressed (well, argued over) in philosophical literature.

Can you expand upon this? It makes no sense as I parse it. Your statement above about whether something we can't distinguish from us without looking at its innards possibly not being conscious is "a dangerous flirtation with dualism" as I see it.

There are two reasons I push the strong AI, mechanistic consciousness view. First it gives us some place to go. A groundwork to begin experimentation and research now, whereas the contrary argument does not.

Second, on a day to day basis - my job - I reduce my conscious thoughts to something that can be done - and subsequently often is done - on a machine. Viz, I am told here is a spreadsheet and here is what info gets phoned in ... we need that information to show up on the invoices and the dispatch board. So I think how I would "do it" all manually, then in stages successively reduce to what machines can do and then write programs to do it with the minimum necessary human intervention. As far as I see, the only real roadblock in principle to doing that with ALL my thoughts is the information complexity problem.

My guess, less than 40 years ... but it could be next week ... or already there in the dark subbasement labs of Google or Oracle with them bogarting the tech until they figure out how to make a dime or two off of it.

I'm also having a really hard time semantically parsing "observer-independent states of ontological subjectivity". Observer-independent and subjective are antonyms.

-- TWZ

I have repeatedly done so. The audience is the same mechanisms of mind that do a primary response to our sensory inputs, with the additional information (that can be regarded in this manner as an additional sensory input) of being internally generated and is "bound" from that sense into the representation of self.

From there the "problem" falls directly into the complexity problem ... figuring out how the mind resolves the "play" and fits the "new" information into its previous data structures and creating new ones "bound" (but the binding is purely syntactic, but of course has semantic reference ultimately) to the old information.

-- TWZ

Yes.

I used no such self referential nonsense.

The idea of self is just another aggregate data structure to the brain-as-computer. The binding of the Cartesian Theater representations of self back to that data structure are instances of pointers and references much as in any computer program, however their biological equivalents are created.

I'm saying that the "binding problem" is another instance people trying to make something trivial sound mysterious.

-- TWZ

Ok we'll go around this circle just one more time.

A "conscious experience" is the process of generating and observing (subconsciously as mechanism) synthesized sensory perception. The process is completely and totally unconscious - mechanistic - start to finish, 360 degrees around any axis you care to identify - apropos as the "cycle" can spin several times a second. We call the synthesis, instantiation, and observation "consciousness", but there is no "consciousness" driving it save for the reality that past iterations of the cycle may cause instantiation of new cycles, or modification of previously pending but not yet "staged" cycles of script, stage, observe, ... script, stage, observe ....

The process is "bound" syntactically (pointers and references) to the data structure representation of self and all related data representations. The entire evolutionary "purpose" is to "game" the relationships between oneself and the outside (available via sensory input) world.

The "not in the dark" nature of the process comes from an impedance mismatch between sensory data association and representation as abstraction. By "staging" pseudo sensory data correlations are revealed to past (real and pseudo) sensory data which have not yet been "abstracted" and integrated into purely unconsciously manipulable information (and I surmise much of the emotive/motivational binding, i.e. to set task priority, comes from the (pseudo) sensory level observation and its correlates with real past sensory experience).

-- TWZ

Staging/Instantiation: (Re)Creation the synthesized sensory material in those parts of the brain where it will be processed as primary sensory data is processed, albeit with the additional "tags" of being internally generated and related to whichever problem set caused it to be generated. (related note on instantiation - see Object Oriented Programming, re instantiating an object).

Scripting: Identifying from the problem set, those external experiences which, with changes should be staged, identifying the stored sensory data necessary for staging arranging it as per the requisites of the problem, and building the instruction data and queuing it to the staging area.

Synthesis: Scripting by commingling and rearranging past sensory data.

Generating: Scripting + Staging.

Observing: Storing real or pseudo sensory input for later reference, processing it to find correlations with past similar inputs and their correlations, and passing it to more abstract processing along with the more abstract correlates associated with past sensory level correlations.

Perception: A portmanteau term for the correlations and abstract relations nearest the sensory input in the processing.

For trivial things we can. For complexity anywhere near ours, we can't ... yet.

You build a data structure.

The same way I build an abstraction of a truck in a computer program. I build a data structure with pointers to all pertinent information about it. (Caveat: I'm well aware that the processing in the brain may not use anything like our computers' data structures, nevertheless as per Turing equivalence they will "mean" the same thing however constructed).

Reveal: You do correlations on new data and find correlations to past experience that was not previously available. See staging and observing above elsewise.

Brown smelly stuff, emo.

-- TWZ

Convention, all caps the real animal. Lower or mixed case the representation - data structure of self.

Because all consciousness does is present a TV set inside your head, on some channels you have control and on others you don't and some in between. Except the reality is that you have no control. It is all done by YOU. And YOU is (for argument's sake) a deterministic mechanism.

You is all that YOU know about YOU that can be expressed as semantic meaning but is data structure and thus syntactically manipulable. YOU know what YOUR consciousness has previously played because it is bound to you as data structure, and can be replayed or re-edited to create a new "conscious experience" at will (i. e deemed useful to YOU) - which of course will also be bound to you. Technically you know nothing of yourself ... but as you are the self representation of YOU, when appropriate "you can express knowledge of yourself" colloqually, which really means that YOU can look at you as data structure, and stage you expressing the knowledge of YOU bound in you to the outside world and then use that representation as a template to act in the real world

You appears "unitary" for two intertwined reasons. All acts of consciousness - that is the theatrical productions are bound to you - YOUR representation of YOURSELF. More damningly, as Dennett has pointed out, you is what YOU know about YOURSELF; you cannot know what you don't know and likewise YOU cannot know what YOU do not know. The "unitary" nature of consciousness is a vacuous necessity.

Nobody does that except as the short form for the following: YOU is a real animal in the real world; you is a real part of the YOU. Both exist and have material correlates and real material consequences in the material world (you being a very large and complex data structure, or rather its bio equivalent, in the brain). The illusion is that you (self) is (completely, or even significantly) YOU. YOU (much less you aka self) do not know everything about YOU; you is only the representation of YOU as agent in and reactor to (via sensory input) in the material world. YOU (nor you) do not know how YOU work internally save what can be inferred by correlating representations of the self. This disparity gives rise to the illusory (instantaneous) and real (forward bound) concept of free will.

Indeed we do. Exactly. And that is and is all of what self IS.

Nobody denies it. Remember heterophenomenology? A complete picture of human cognition must explain why people think they are doing what they think they are doing in their own heads. It's complex and subtle but this does that in rough outline.

-- TWZ

Yes.

Incorrect. Your remembered experiences ARE the correlations of (pseudo) sensory data (which can be replayed as useful in quest of further correlations). "Having an experience" is the process of creating such correlations.

-- TWZ

And my hypothesis is that the synthesis, instantiation, and observation of pseudo sensory data (as a "computational" model, which Edelman disdains) is the C* process.

With the net result of explaining why C is all pseudo-sensory and why C and C* remain synchronized ... all a result of

s is sensory data from various channels (including internal and emotive state data) at any one instant

S* = M is a convoluted vector of mind state at any moment.

S0 = OnePermutation(S*)
C0 = C*(S0, M)
Ct = C*(, M) (x, y, z, t are unique)

-- TWZ

Quite. Both Dennett and Edelman note Damasio's work. In my model I posit emotiive and physiological state as "sensory" input. These are corrielated at any point with other real and/or pseudo sensory data to evaluate and react (build/modify/reenforce data structures and ultimately decide upon action). It wasn't part of my original model, and I don't think Dennett mentions it in this context, but from what I've read of Damasio, I would suspect that the motivation (setting task priority and establishing connection weights, in computer terms) is established largely at this level.

The motivational binding may have more to do with the "not in the dark" nature of conscious experience than the re-valuation due to lack of storage space aspects of re-imaging past sensory data in consciousness. Physiological reactions take significant time (fractions of seconds to seconds) to establish where "syntactic" (completely unconscious) processing takes relatively milliseconds (and is likely to be massively parallel).

What I think is happening is that our "wiring" for various physiological states remains intact and thus through correlation we quickly do "recognise" ourselves internally as the same stimulae produce what is internally (though probably unique for each individual) similar physiological responses that are presented to the brain hardware as sensory data in substantially the same way.

Roughly, yes I agree. Are there cases where people have had extensive brain surgery who assert that they don't recognise themselves? That they are not the same person? I seem to recall reading some of such.

-- TWZ

More to come on other theads!

-- TWZ