Reflections in a Chessboard
Posted on 03 September 2010
“Consider the Chinese Room, or better still consider Deep Blue, the chess machine. Nobody claims that Deep Blue has consciousness, but it has intelligence in the very narrow sense of playing excellent chess.”
And Grandfather Clock has intelligence in the very narrow sense of being able to count the minutes and hours correctly, adding up the sums in its head, and telling me the correct time, by deciding to play the chimes hanging in his case. Oddly, Grandfather Clock always decides just exactly on the hour and half-hour to ring the correct chimes. I am astonished at how accurately Grandfather Clock’s sense of timing is, how tirelessly he attends to his task, and how he never loses count or mistakes the number of minutes in an hour. No doubt Grandfather Clock is helped in the tireless precision of his thinking process by the wheels and gears that make up his brain. Nonetheless, we all must commend Grandfather Clock for his diligence and uncomplaining attention to detail. He is as patient and devout as a Beefeater Guard who stands before Buckingham Palace, and, like them, he never stirs from the spot where he has decided to stand.
I am kidding, of course. Grandfather Clock is a machine.
It does not have any intentions, diligence, ability to count, ability to add numbers up to 60, or any other mental operations of any kind. It is not even alive. It is a machine. Deep Blue is also a machine.
If you took the time to write down every possible combination of chessman positions on a fat deck of cards, and a chessmaster took the time, for each and every possible location of chessman, to write down what he thought was the best move on the back of the card, you could pretend to play a game by setting up the board, finding the card that represented an opening, making a response, and finding the card representing the new position of the chessmen, and looking on the back of that card to see what the chessmaster thought the best response would be in that situation.
However, the deck of cards would not be playing chess. The deck of cards would not have any intention, any awareness of the chessmen or any idea of the relation between the chessmen and what they symbolically represent, or what moves or victory conditions the game entailed. The deck of cards would not be aware of the game at all. Its a deck of cards.
The deck of cards cannot win by concentrating better of by thinking of an new strategy or by correctly weighing the options open to it or by correctly guessing which pieces you are willing to sacrifice. It cannot concentrate of think or weigh options or guess.
The deck of cards, in this thought experiment, were all written out beforehand by a chessmaster. The chessmaster can think, because he is a human being with a rational soul, and an ability to make up things that do not exist in real life, such as the rules to games.
If you asked Deep Blue to play Chancellor Chess, or to abide by a house rule that forbids en passant capture of pawns that moved twice on their first move, you would not get an answer, because you cannot ask Deep Blue to change the rules. Likewise you cannot ask a deck of cards to change the rules. These things do not understand what “rules” are or why we have rules in games. They do not understand what a game is, because they are not alive and do not think.
The word “intelligence” is being used here in what is actually merely a misleading metaphor. The intelligence of Deep Blue is in the programmer of Deep Blue, who was not someone unaware of the rules of chess.
It is as if we said a red and stormy sunset was an “angry” sunset and as if next, without realizing it is a metaphor, we sat and talked quite seriously about whether the Man in the Moon had offended the sunset to make him angry, or whether merely an upset stomach or colic had angered the sunset.
The error here is the informal fallacy called ambiguity: merely using a word that has two meanings as if it had one meaning. The intelligence (meaning 1) displayed by a clockwork, a deck of chessmove cards with instructions on them, or a computer program, is the intelligence (meaning 2) of the watchmaker, chassmaster or programmer.
Also, there is a formal error in logic here, called the fallacy of the undistributed middle. The syllogism “(1) Men’s thoughts can be described in terms of complex decisions or formalization (2) Computer program logic can be described in terms of complex decisions or or formalization therefore (3) Men’s thought are computer program logic.” suffers the same error as the syllogism, “(1) Englishmen are men (2) Frenchmen are men therefore (3) Englishmen are Frenchmen.”
A more pointed example is “(1) Lincoln at Gettysburg recited a speech (2) My phonograph record recited a speech therefore (3) Lincoln at Gettysburg is the same as a photograph record.”
The whole materialist argument is based on a mildly surrealistic premise that the only thing we know about anything are its empirical properties, and that ergo any non-empirical properties causing those empirical properties must ergo be the same in every way. I call it surreal because we are dealing with a case where the chain of cause and effect is known to be different. As if someone were to argue that, because my wife loves me, my wife’s image in the looking glass must also love me, on the grounds that she has the same look and smile and expression in her eyes. However, the smile of the image in the glass is caused by photons bouncing off the wife, not by any affection or admiration or lust or devotion on the part of the looking glass. It’s glass. Not only is it not loving, it is not even alive. The smile on the real wife is caused by the wife’s soul, or the great god Cupid, or her bad judgment or however you want to say it. The glass is not even really smiling because it is an image of a smile, not a smile.
Likewise here. Clocks do not know math, and do not know the number line, and do not count numbers and match them to passing minutes and hours. Clockmakers know how to count. Clockmakers make a machine that reflects this the way a looking glass reflects an image. We only say a clock “tells” time because we, the intelligent beings, tell time using the clock as an instrument, and it is no more than a convenient shorthand of speak to attribute to the tool the mental operations we perform using the tool.
Likewise, Deep Blue does not play chess or wins games any more that he feels satisfaction over games well played or broods over mistakes he made. He is not a he.
Deep Blue is an image, a complex one, but no different than a clock. We only say that Deep Blue “plays chess” as a convenient shorthand of speech because we, the intelligent beings that play chess, managed to write down chess moves and groups of chess moves in a hierarchy, so that we, using the machine as a tool, can look up what move might beat the move we just made. Deep Blue is not “doing” anything. The chessplayer is looking up a move in Deep Blue’s database much as if he were looking up a move written on a page in a book in a library, but he is using Deep Blue as a mechanism to look up the move.
Another example:
f it should turn out that the live speaker I thought was Mr. Lincoln is a Disneyland automaton after all, this does not lead logically to the conclusion that I am also a automaton: it leads to the conclusion that the real Mr. Lincoln, a living soul, wrote a speech, and that a real Mr. Disney, another living soul, figured out how to combine a phonograph record with a clockwork manikin to create an image—and this image is as false an illusion as shadows in a looking glass.
It would be surreality more surreal if anyone were to propose the argument that because we cannot tell the difference, using only our eyes, between a wife that loves us and a looking glass that loves us, therefore our wives are merely looking glasses, perhaps more complicated, but otherwise nothing different from flat reflections in a surface.
Likewise here, there is more than a little surreality in the argument is that since we intelligence beings can (after much painful thought and careful work) embody a reflection of our intelligence in a clockwork, ergo all intelligence is clockwork, including our own, therefore painful thought and careful work do not actually exist.
And when you point out that clockworks do not think and are not alive, and do not have any personal selfhood and do not have any point of view or any ability to make decisions or any subjective sensation and cannot feel pain or feel desire or feel the impulse to act, the answer you get back is …
Well, I don’t know what answer you get back. No radical materialist with whom I have ever spoken as given an answer. I ask the question, and they change the subject, and act as if I have asked nothing. It is kind of eerie.
Since this point is often lost, allow me to dwell on it for a moment:
The argument for radical materialism is based on the unreal premise that, because our ear cannot tell the difference between a living speaker and a good recording, therefore living speakers are nothing but recordings. They have grooves in their brain similar to the needle grooves in a phonograph, and something like a phonograph needle sends a nerve impulse to muscles in the throat, which makes the words come out.
Well, yes, as far as the ear is concerned, you cannot tell the difference between a phonograph record and a live speaker, if you are a moron or a philosopher deliberately ignoring all the surrounding facts. Because in addition to the ear, I am also aware of the fact that when I talk I am making up the words I say and deciding what words to use; and when a phonograph record is being recorded, that process of thought is also present in the speaker making the recording, not in the record disk. The words of a live speaker are like the smile of the wife: it happens because she wants to smile. This is the immediate cause. The words from a phonographic recording are like the smile of an image in a mirror: it happens because photons reflect off a surface evenly enough to create an illusionary image or shadow in the eye. The smile of the wife is the remote cause, the reflection from the glass is the immediate cause. An argument that says, “As far as your eye is concerned, you cannot tell the difference between an image in a looking glass and a real woman, therefore women are images in mirrors” is not just bizarre, it also makes formal and informal logical errors, including the error of not making any distinction between immediate and ultimate causes.
Drawing this analogy back to the case at hand, what we have here is the argument that Deep Blue is “intelligent” in the limited sense that it can play chess well enough to beat a chessmaster. But of course the programmers and chessmasters who gathered the tens of thousands of man-hours of experience with chess written into the various electronic files of the machine like a library of chessgames are ignored with the same blithe indifference that the real woman creating or causing the image is ignored in the argument that mirrors are as real to your eye as reality.
For those of you who like letters, let us call the non-empirical but ultimate cause of a deliberate act the letter “A”, and let us call an instrument that can reflect or mimic such acts (including a clockwork, a deck of chessmove cards, a mirror, or a computer) the letter “B”, and let us call the end result we see with the eye or hear with the ear (including a faithful report of the time, a smile of love, or a chess move) the letter “C”.
In one case A causes B causes C, where “B” here is something that reflects. B is our Grandfather Clock, our looking glass, our Deep Blue. In the other case A causes C, where “A” is the watchmaker, the wife, the chessmaster. In both cases, “C” is some material thing we can perceive with our senses, eyes and ears. In both cases, “A” is something we cannot perceive with our senses, the thoughts in our consciousnesses.
We have two cases, one where a living soul does something immediately, and one where a living soul uses an instrument to do something mediately.
An example of immediate acts includes me speaking my dying words aloud to someone in earshot, and I tell him what I want done with my property after I die; an example of a mediate act is when I use an instrument or medium as an intermediary, such as when I write the same words in a letter or a Last Will and Testament, and a person outside earshot later reads them. Speaking aloud is A causes C. Writing a will is A causes B causes C .”A” is my intetion, where “B” is the piece of paper on which the will is written, and “C” is when my words are understood.
The argument of the radical materialist boils down to the empirical proposition that we must abide by the rule which says we are not to speculate about things our sense cannot grasp: we are not allowed to say that “A” causes anything, because things like thoughts and intentions have no size nor shape nor weight, therefore we must pretend to take no notice of them.
The first mistake of the radical materialist is that what empiricism ignorex does not exist. “A” does not exist.
The second mistake is to conflate mediate and intermediate causation: the first case (A causes B causes C) is treated as the second case (B causes C).
The third mistake is merely the fallacy of the undistributed middle. (1) A causes C (2) B causes C therefore (3) A is B.
I am not sure if it counts as a separate mistake to reintroduce “A” into the discussion after we strict empiricists all agreed not to make any speculations about thoughts or intentions or anything else we could not see nor touch.
The radical materialist misinterprets the first case, and treats it as if B were the sufficient and sole and ultimate cause of C. He then makes the fallacy of the excluded middle: (1) B causes C (2) A causes C therefore (c) A is B.
For those of you who don’t like letters, let me rephrase the paragraphs above. The argument of the radical materialist boils down to this: Since the only thing we can really know about a woman’s smile or the Gettysburg Address is what we see with our eye and hear with our ear, and since we cannot tell the difference between a mirror and a wife, or between Lincoln alive and Lincoln on a phonograph (if you will forgive the anachronism) therefore we must treat images as if they are reality, and therefore images are reality; therefore we must treat proximate causes as if they are remote causes, therefore there are no remote causes.
The argument forgets one thing. Since I know what it is like from my point of view when I smile, and since I know what it is like from my point of view when I make a speech or make a chessmove, and since other people are real, live people and not manikins or images in a mirror, therefore I know that other people make speeches and smile smiles and make chessmoves for the same reasons I do.
Since in my case, me being aware of my own consciousness, I have the privileged information of knowing that A causes B causes C: in a case where, limited by my senses, all I see of other people is that B causes C, I am forced to assume that there is some cause A present there as well.
Suppose I am in a cellblock. I can see the prisoner’s feeding trays slid by the guard into a slot in the solid cell doors, and emerge later with the food gone. The only cell I can see into, however, is mine. I know I eat my food; but I do not know if the other prisoners are eating their food or dumping it on the floor. Indeed, I do not even know whether the prisoner in the cell at the end of the row has not made a cunning automatic machinery out of the slats of his cot and strips torn from his striped uniform that takes the proffered tray from the slot and dumps the food on the floor, and returns the tray to the guard.
Let us further suppose that I am a criminal mastermind able to make such an automatic machine myself, so I know that it is possible to dump the tray on the floor and return it clean through the slot without eating the food.
But I do know that since I eat my food (“A”) that I am the cause of the empty tray returned through the slot in the door (“C”). I also know that it is possible for some intermediate cause (“B”) such as a cunning automatic machine made out of cot slats to return the clean tray through the slot to the guard without eating.
It is feasible, in this situation, for me to conclude that therefore my brain is a cunning automatic machine made of cot slats? Is it feasible for me to conclude that eating the food is the same as dumping it on the floor? It is feasible for me to conclude that eating the food is an epiphenomenon or an illusion and that the real reality, the only reality, is that my brain is a machine made of cot slats and that I never actually ate any food?
I would say that is not a feasible conclusion. If so, why should this unfeasible and frankly absurd argument somehow become worthy of consideration if instead of clean trays, we talk about chessmoves, and instead of eating, we talk about pondering and devising chessmoves based on a knowledge of the rules of chess, and if instead of a cunning machine made of cot slats, we talk about Deep Blue?
Admittedly, I can be fooled into thinking Lex Luthor is still in the cell next to mine, merely because I see his clean tray being returned through the slot in the door after mess time. Maybe he is in the cell and ate his food, or maybe he escaped, leaving behind a cunning machine made of cot slats to fool the guard. But the mere fact that I can be fooled is no grounds for me to conclude that Lex Luthor’s brain is the same as a machine that he makes. If conclude Lex is still in his cell when he is fled, my conclusion is what we call a mistake. I am wrong. My inability not to be fooled into mistaking people from machinery does not mean people are machinery. My inability to peer into the other cells in the cell block does not mean all other cells have always been and must always be empty, occupied by machinery, not inmates.
I submit the logical conclusion is this. If I see an empty tray being return through the cell door slot after mess time (1) the next cell either has an inmate eating the tray of food or a cunning machine made of cot slats dumping the tray on the floor (2) if there is a cunning machine made of cot slats, it was made by an inmate who has since escaped.
Likewise, if I see an image in a mirror, or a chessmove written on a card deck of instructions, or hear a clock chime the time, or see a computer react to a command by looking up a chessmove and presenting it to me on an electronic screen, all these things are intelligent actions but not intelligent actors. They are not actions at all, but reflections, reactions. They react intelligently due to the intelligence of the wife, the chessmaster, or the watchmaker, and this intelligence is the unseen ultimate cause of the intelligent actions I see.
The difference between a machine and a person is that the machine functions for another and a person functions for himself. Also, the parts of a machine have no necessary relationship to one another and do not combine by nature, only by artifice.
+ + +
“Intelligence” comes from Latin: inter legere, “to read between (the lines).” That is, to grasp information that is not physically in the data.
I lost the ability to think of artificial intelligence as being real intelligence shortly after I learned how to program a computer. The more I learned about programming, the more I realized that a computer always and only does exactly what you tell it to. Nothing more. If the computer does something I did not intend, it’s because I didn’t completely think through what I told it to do. If the unintended behavior is unwanted, I call it a “bug”. If I think that the unintended behavior is acceptable, I call it a “feature”. If I don’t understand why my computer program is doing something, I can always just trace the code, looking at each line, and how the values of the variables are being manipulated. When I find a line that does not have the desired affect, I can correct it to do what I want. Taking programming classes ruined my ability to enjoy Star Trek, or any science fiction that involves robots or AI. I had to re-learn to suspend disbelief.
I think the temptation towards radical materialism comes from looking at neurons as lines of code. With sophisticated machinery, I can look at the brain and its inputs, and predict the outputs based on which neurons are stimulated, just like I can run a stack trace on my program and see what the outputs were for any given inputs. We tend to group things that behave in similar manner as similar things, so we surmise that since brain inputs and outputs can be detected and predicted like a computer, maybe the brain is just a sophisticated computer. A logical fallacy to assume this is true, as you pointed out, but a disorienting consideration none-the-less. (You once mentioned, I recall, that good SF should produce this disorientation, like that Copernicus might have felt upon realizing we aren’t at the center of the universe, or that Darwin might have felt at realizing that each species wasn’t individually created.)
But isn’t it worth at least considering the idea of brain as a computer? What if all my thoughts aren’t really creative like I think them to be, but simply the results of lines of code–predictable outputs for given inputs? Maybe my mind is just a finite state machine, and poetry isn’t really poetry, but algorithmically produced gibberish in which I have been programmed to find meaning? The implications, of course, are staggering, and it would mean there’s no such thing as morality, love, or justice. For that reason alone I reject it (or have been programmed to reject it!), but it does produce that interesting SF-like state of disorientation just to consider the possibility.
The other reason, I think that materialism is tempting is that proper philosophy is no longer taught. Even as a graduate of a liberal arts college, I still did not know or understand many of the basic concepts until I started reading yours and Mr. Flynn’s blogs.
I was once explaining programming to a child by saying that a computer is dumber than an earthworm. An earthworm can figure out that it has to go around obstacles, but a computer has to be told.
Oh dear, problems submitting the comment. I hope I’m not duplicating this.
This is, no doubt, true of all programs you have encountered; but it’s not necessarily true of all possible programs. There comes a point of complexity where you are no longer able to keep all the variables in your head, and then, when asked why the program did X instead of Y, you would be unable to answer. And I further suggest that all useful programs are like this to some extent; because if you knew what the output would be, why bother to run the program?
Deep Blue most certainly can play Chancellor Chess, or no-en-passant chess; you merely need to change the algorithm where it checks what is a legal move. If we have neglected to supply Deep Blue with sound sensors and natural-language-processors, so that this change must be made by means of electrical currents rather than speaking out loud, that is a side effect of our bad design, and no reflection on Deep Blue. As for the other things you say Deep Blue cannot do, I didn’t claim it could. I offered it as a very narrow example of intelligence, the ability to do tasks, without consciousness, the awareness that one exists. If I say “Deep Blue can play chess”, it is no counterargument to say “But it cannot invent the game of chess”; for I never claimed it could.
The question is not whether any particular existing machine has the full flexibility of a human; we know perfectly well it’s not so. The question is whether such a machine can in principle be built. And then I refer you again to the quark-level simulation. Let me go that one better, in fact. I will build myself a supercomputer capable of solving the QFT equations for something the size of a human brain; then I shall insert it in a human skull and carefully solder the metal inputs to the nerves. Thus, when a signal travels up the human’s arm saying “Hot”, this will at some point be translated to an electrical voltage within my supercomputer, which will simulate the response of the brain (now classified as medical waste; a minor sacrifice FOR SCIENCE) and eventually re-translate its internal voltages into a nerve signal jerking the arm away.
Now, pray attend carefully to what I say this machine will do. I do not claim that it is conscious; it seems to me that consciousness arises from particular patterns of energy-and-matter flow, and the flows within my computer will be very different. I do, however, claim that this machine will act the same as its human predecessor. (Yes, I know that you don’t like to use the word ‘act’ of machinery. Pray pardon the clumsiness of the English language; I can think of no better word.) If its wife leaves it, water will come from its eyes. It will write bad poetry, or possibly even good poetry. And a human who did not know that the source was a robot, would be moved or not moved by that poetry just as he would be moved or not moved by that of the human predecessor’s.
Again, please distinguish carefully between what I claim, and what you think a materialist ought to claim! If you wish to say that the meaning of the robot poetry does not arise until a human reads it, I make no objection for now. If you say that the robot does not understand its chess game, or take any satisfaction in winning, or derive pleasure from sleeping with its wife, that is all fine for the time being. What I claim, however, is that all the external behaviour of a human can be formalised in this manner. This robot would be able to learn French by the usual human method of hearing it spoken and seeing it written. If you asked it to play a chess variant, it would be able to do so, or then again it might refuse if the original human would have found your variant uninteresting. I defy you to think of any act which a human can do, which this robot cannot, that is not internal to the robot’s state of mind (or lack of one).
There comes a point of complexity where you are no longer able to keep all the variables in your head, and then, when asked why the program did X instead of Y, you would be unable to answer.
And it would be different from person to person.
Would that mean its ontological status differs from person to person?
No. And therefore, when you find that a human’s actions are predictable in principle but not in practice, what ought you to conclude about the ontological status of human brains?
“When”
What question begging in that word. . . .
Are you claiming, then, that the atoms within a human brain do not obey the usual laws of physics? Is their energy not conserved? I do not ask as a criticism, but merely to clarify.
Ah, the question begging in the questions you ask to clarify.
Well, which is it? Do the atoms in human brains obey the known laws of physics, or do you believe there are other laws? Or is there a third alternative?
Atoms in human brains obey the laws of physics.
I wish to use intelligence to refer to apparently-from-the-outside purposeful behaviour, and consciousness to refer to apparently-from-the-inside purposeful behaviour. Even if you dislike the usage, can we not agree that it is a useful one for discussing the operations of machines that imitate human action?
I wonder if this doesn’t muddle things more than it clarifies them. I’m not at all sure you can get intelligence without consciousness. That’s certainly the case for computers; their purposeful behavior is the result of our consciousness. I would argue, and I think John would too, that our purposeful behavior, and that of ants, comes from the consciousness of the Creator.
“I wish to use intelligence to refer to apparently-from-the-outside purposeful behaviour, and consciousness to refer to apparently-from-the-inside purposeful behaviour. Even if you dislike the usage, can we not agree that it is a useful one for discussing the operations of machines that imitate human action?”
The argument here is based on an ambiguity. What you call consciousness, what I call intelligence, produces what you call intelligence, what I call intelligent action. A causes B. Your argument rests entirely on pointing at examples of B, tools such as Deep Blue, or hypothetical robots, or hypothetical Chinese Rooms, and studiously ignoring the cause of B, which is A, such as chessmasters who programmed Deep Blue, Susan Calvin who designs robots, or Chinamen who went to the trouble of reducing their entirely language to an algorithm and filling up a Chinese Room with indexes and libraries.
The argument then goes that since B arises without A, therefore inside us humans, there is only intelligent action B, but no intelligence A. Since humans can build clocks that think like human, all human thought is clockwork. As best I can tell, these are merely disconnected statements with no logical relation to each other, and no empirical relation to reality.
The terminology is misleading and I strongly suggest we not use words to refer to things that are the opposite of what the words normally mean. Let us not use the word “intelligence” to refer to the automatic actions of clockworks and unpiloted steam engines when the intelligence of those tools rests in the tool maker who made them.
Of course there’s a third. Your question-begging, however, leads me to believe I couldn’t explain it to you no matter how I tried.
That may of course be the case, either through my obtuseness or the actual incoherence of your alternative; but if you are not willing to have your theories exposed to public view, that is your privilege. However, may I suggest that in that case you are not entitled to mock and carp? “Your argument is stupid, and I have a knowckdown counter to it which I won’t explain because you’re too stupid to understand it” is not, I think, quite the impression you want to give. It is not really the best possible tone.
“No. And therefore, when you find that a human’s actions are predictable in principle but not in practice, what ought you to conclude about the ontological status of human brains?”
I conclude that “predictability” is not relevant to the question of whether human action cannot be described except in terms of final causes. My wife can predict with uncanny accuracy at what time I wake up in the morning and what I will eat for lunch today; whereas I cannot predict the motion of a pendulum swinging between a triangle of magnets: and yet, it is clear enough that I make decisions and have free will, and that a pendulum does not.
All right, and the robot that simulates your brain, does it have free will? In any given circumstance it acts exactly as you would. (Noting in passing that Mary, if I understand her correctly, does not agree with this.) Yet it seems to me that if we wished to describe your actions, we could consult the robot and say, “It brushed its teeth because this register contained the value 0xBAD7EEF”, or some rather more long-winded statement in terms of plain electronics; and we could also say “It brushed its teeth because, in the same circumstance, Mr Wright would have worried that his breath was bad”. These explanations seem equal in power; the second, clearly, is a useful shorthand, but otherwise there does not seem to be any reason to prefer it.
I suppose I do not understand what you mean by free will. We have established that it is not a question of predictability or its lack, either from quantum mechanics, chaos theory, or any other cause; you say, if I understand you correctly, that you would have free will even if the atoms of your brain were classical objects following plain Newtonian trajectories. Is it self-awareness, then? Moral significance? I have seen you use both these terms in your explanations of what you mean by free will; yet they don’t seem precisely equivalent to what I usually understand by free will.
I do not know whether we are actually disagreeing on anything at this point. It may be that we have the same view of the universe, but use different words to describe it, and have been led astray by a mere divergence in our technical vocabularies. My advisor describes this situation as “agreeing violently”; it arises in physics discussions more often than you’d think.
I have found that atheists and theists sometimes end up talking past each other when discussing the issue of free will. Let me offer two definitions. “Type 1 free will” deals with the freedom of the will in relation to the laws of physics. John says a pendulum isn’t free because it has no choice but to obey the laws of physics. Well, either the mind obeys the laws of physics, or it does not. If it does, then is it free? If so, is this freedom based upon quantum randomness? If so, what are the implications of this? If it does not obey the laws of physics, well… I can’t add much, because I don’t adhere to this view.
“Type 2 free will” deals with the freedom of the will in relation to God. Does God interfere with man’s will? Calvinists would say “yes, as concerning the choice to believe”; non-Calvinists would say no.
“Well, either the mind obeys the laws of physics, or it does not. ”
It clearly does not. The laws of physics describe events in terms of mass, length, duration, candlepower, moles of amount, temperature, amperage, or some combination thereof. The laws of physics deal with material events only according to mechanical causes only. A description of the position of brain atoms in the brain or a description of ink molecules on a page does not tell you whether what is written there is true or false, efficient or inefficient, just or unjust, sane or mad, good or bad, beautiful or ugly, happy or sad.
There is a certain special category of words we use to describe the relationship between symbols and the thing the symbol represents: true and false, accurate and inaccurate, logic and illogical, meaningful and meaningless. This category of words has to do with representation. Representation is when in the mind of an observer, on things “stands for” another thing.
Now, it should be patently obvious that a physical object cannot “stand for” another physical object in the absence of an observer who makes the purely mental and non-physical association between those two ideas in his mind. A book written in the language of the long dead Martian race is not a book: it is merely marks without meaning on a sheet of iridium. It does not become a book until some creature with a mind from another planet lands and translates the meaning of the symbols. But the “meaning” and the “symbols” are not material properties of the sheet of iridium.
The words in that category cannot possibly be reduced to measurements of quantities of mass, length, duration, candlepower, moles of amount, temperature, amperage. This is because describing the chemical properties or position of ink marks on a page or the electronic properties of nerve firing in the brain do not tell you anything about the non-material properties those marks and those nerve firings represent.
Since human behavior, even if it is entirely determined by iron fate, cannot possibly be described except in terms of final causes (what the man meant to do) and since abstract thought cannot possibly be described or defined except in terms of symbols.
I am not saying that men are somehow immune from the laws of physics. I am saying the laws of physics are ways of describing something, and the laws of physics do not and cannot in this or any other universe ever describe human action.
[wrf3] “Well, either the mind obeys the laws of physics, or it does not.”
[John] “It clearly does not. The laws of physics describe events in terms of mass, length, duration, candlepower, moles of amount, temperature, amperage, or some combination thereof. The laws of physics deal with material events only according to mechanical causes only.
But those mechanical causes include quantum, and other, forms of randomness. Those laws of physics not only can encode, but can enable algorithms to act on data. Couple that with randomness and you have the basis for what we call intelligent behavior.
There is a certain special category of words we use to describe the relationship between symbols and the thing the symbol represents: true and false, accurate and inaccurate, logic and illogical, meaningful and meaningless. This category of words has to do with representation. Representation is when in the mind of an observer, on things “stands for” another thing.
And I argued, in my post on 9/4 @ 5:19 that there are two non-physical objects for which the representation of the thing is the thing: intelligence (as evidenced by the Turing test) and God (as evidenced by the doctrine of the Trinity: “and the Word was God”. That is, “what God says” and “what God is” are the same thing). You keep citing physical things for which the representation clearly is not the thing. I’ve offered two things where that distinction does not hold.
Since human behavior, even if it is entirely determined by iron fate, cannot possibly be described except in terms of final causes (what the man meant to do) and since abstract thought cannot possibly be described or defined except in terms of symbols.
But software can; and thought is (complex) software.
But those mechanical causes include quantum, and other, forms of randomness.
Ah, the magical quantum invocation. But a great deal depends on which of the many different quantum theories is correct.
And randomness is a description of our ignorance, not a cause of anything. A random pattern, such as the Weibull or the 3-parameter lognormal, is just that: a pattern. It is analogous to the human soul in being the form of something substantial. But how can pure form be a cause of anything.
That which we usually call “random” is typically the intersection of two causal chains, each link of which is determined by a natural law or the intention of an actor. (In which case, the natural laws can also be understood as the Intentions of an Actor.) The fact that two causal lines intersect does not make a causal relationship between the thing on one line and the thing on the other. It is our insistence on seeing significance in concatenation that muddies the waters.
Ah, the magical quantum invocation.
Label it what you will, whether “magical” or “mysterious”, it’s the way the universe works.
But a great deal depends on which of the many different quantum theories is correct.
There’s one quantum theory, but several interpretations of that theory. The formulas that describe what we observe are the same.
And randomness is a description of our ignorance, not a cause of anything.
That’s simply not true. As Donald Knuth (a patron saint of Computer Science) wrote, “Indeed, computer scientists have proved that certain important computational tasks can be done much more efficiently with random numbers than they could possibly ever be done by deterministic procedure. Many of today’s best computational algorithms, like methods for searching the internet, are based on randomization. If Einstein’s assertion were true, God would be prohibited from using the most powerful methods.”
‘ “(quoting me) abstract thought cannot possibly be described or defined except in terms of symbols.”
But software can.’
Please provide me an example. Show me some software that does not make reference to final cause, that is, software not “for” something that does not “represent” something. Define a line of software merely in terms of quantitative measurement of mass, length, duration, or other physical properties without reference to the language or other non-physical symbolic correspondence that language contains but that empirical objects do not?
“…. and thought is (complex) software”
Allow me to quote Thomas Hobbes, who agrees with you: For seeing life is but a motion of limbs, the beginning whereof is in some principal part within, why may we not say that all automata (engines that move themselves by springs and wheels as doth a watch) have an artificial life? For what is the heart, but a spring; and the nerves, but so many strings; and the joints, but so many wheels, giving motion to the whole body, such as was intended by the Artificer?”
Now then, is Mr. Hobbes speaking literally, or is he employing a misleading metaphor? When you call your own thoughts software (that is, you are saying your thoughts are unaware and unalive lines of electrons having a symbolic meaning only to an intelligence outside observer but not to you, moved by external forces toward the result intended by the designer) — are you employing a metaphor like that of Mr. Hobbes, who calls your heart a mainspring and your life merely a jerking of the limbs, are do you mean to be taken literally?
“(quoting me) abstract thought cannot possibly be described or defined except in terms of symbols.”
[wrf3] But software can.
… Define a line of software merely in terms of quantitative measurement of mass, length, duration, or other physical properties without reference to the language or other non-physical symbolic correspondence that language contains but that empirical objects do not?
It’s just 0′s and 1′s in the right pattern and all digital logic can be represented by NAND (or NOR) gates wired in the proper sequence. Do you want me to give you the NAND gates that are equivalent to (INCF X)? Or the NAND gates that are equivalent to a LISP compiler? We build abstraction on top of abstraction on top of abstraction, but it’s just NAND gates connected the right way. You’ve built your argument on the claim that the representation of the thing is not the thing, but I’ve given you two counter-examples, which you haven’t yet addressed.
[wrf3] “…. and thought is (complex) software”
When you call your own thoughts software (that is, you are saying your thoughts are unaware and unalive lines of electrons having a symbolic meaning only to an intelligence outside observer but not to you, moved by external forces toward the result intended by the designer)But I don’t say that the symbolic meaning is only to an outside observer. Meaning arises out of the connections between things and we are able to create our own meaning. We are also able to share it and so build up a language of common meaning.
— are you employing a metaphor like that of Mr. Hobbes, who calls your heart a mainspring and your life merely a jerking of the limbs, are do you mean to be taken literally?
I’m not employing a metaphor. As I’ve said several times now, there are at least two things where the representation of the thing is the thing: God and intelligence. Software isn’t any different, really, from Logos. Scripture says, “In Him we live and move and have our being.” Certainly He is the source of Logos. What we’re arguing over is whether or not we can duplicate it.
Good question. My answer is here
and the robot that simulates your brain, does it have free will? In any given circumstance it acts exactly as you would.
and the reflection in the mirror, does it have free will? In any given circumstance it acts exactly as you would, except in reverse.
““It brushed its teeth because this register contained the value 0xBAD7EEF”, or some rather more long-winded statement in terms of plain electronics; and we could also say “It brushed its teeth because, in the same circumstance, Mr Wright would have worried that his breath was bad”. These explanations seem equal in power; the second, clearly, is a useful shorthand, but otherwise there does not seem to be any reason to prefer it. ”
The two explanations differ in nature. One is a description of mechanical cause, and tells you only the mechanical cause and not the meaning of the action; the other is a description of final cause, and tell you the purpose and meaning of the action, omitting the mechanical cause.
This whole weird argument seems to be based on your assumption that these two types of explanation are (1) of the same nature and (2) mutually exclusive; my whole argument is based on the assumption that they are (1) of entirely different natures and (2) mutually independent or indifferent.
By indifferent I mean that a false statement in one type of explanation would not necessarily make the statement in the other type of explanation false. For example, if the Wrightbot tech found out, to his chagrin, that it was value value 0zBAD7EEF not value value 0xBAD7EEF that make the muscles of my arm move and pick up the toothbrush, it would not necessarily be the case that I, from the inside, was not trying to brush my teeth to avoid bad breath. Likewise, if upon cross examination, the jury became convinced that my true motive for scrubbing my teeth was not because I feared bad breath but because I had some other motive entirely, such as a desire to tarry in the bathroom an extra moment or two, it would not necessarily be the case that value 0xBAD7EEF was not the one making my nerves and muscles go through the tooth-brush picking up subroutine.
Since this is a rather difficult and subtle distinction, I ask if you understand it? I am not asking if you agree or disagree, I merely ask if you know what the words mean when I talk about mechanical cause and final cause?
I understand the distinction, yes, although I consider it to be ultimately a false one.
It seems to me that Mr Wright addressed this here:
If (I) conclude Lex is still in his cell when he is fled, my conclusion is what we call a mistake. I am wrong.
Yes, you may in some alternate reality be able to build a machine to simulate consciousness. But that machine is not conscious.
Apologies if I misunderstand.
I believe you did misunderstand me. Let me draw your attention to this sentence: “I do not claim that [the robot] is conscious”. I don’t say I would simulate consciousness – indeed this looks like a contradiction in terms; I might be able to build a machine (or an artificial lifeform, if you prefer) with consciousness, but not with simulated consciousness. Either it has subjective experience or it doesn’t! What I do claim, however, is that this robot would be intelligent; that is to say, it would act with apparent purpose and would change the world around it in a non-random manner. It is precisely to distinguish between intelligence and consciousness that I am introducing the example. Intelligence is to act in ways that appear purposeful, with a purpose not externally imposed, to an outside observer; in this sense an anthill has intelligence. (A clock does not, because the purpose is external.) Consciousness is to act in ways that appears purposeful to an internal observer, or rather, it is to have an internal observer in the first place.
I think it useful to distinguish between these two concepts, although in most cases we encounter them together, since we usually interact with other humans, who are both intelligent (more or less!) and conscious. However, even in nature we can point to examples where they are separate; a dog is conscious but not intelligent, at least not compared to a man; an anthill is (slightly) intelligent but not (as far as I can tell) conscious.
I see. Thanks for clarifying.
“What I do claim, however, is that this robot would be intelligent; that is to say, it would act with apparent purpose and would change the world around it in a non-random manner.”
Likewise, you can claim that a shovel digs holes for a fencepost, because the shovel craves a fence. The statement only makes sense if we ignore the handyman holding the shovel.
Quite so, but the robot does not require any such outside impetus, any more than you do. It acts entirely on its own. If every human in the world were replaced by such robots, human civilisation would continue, although nobody would now be experiencing its delights. It seems to me that there is a qualitative distinction between such objects, and a shovel.
“Quite so, but the robot does not require any such outside impetus, any more than you do. It acts entirely on its own.”
Then it is not a robot. The word “robot” means a made thing, something programmed.
“It seems to me that there is a qualitative distinction between such objects, and a shovel.”
It seems to me that there is not. Perhaps you are thinking of the Tin Woodman of Oz, and not actually considering what a robot actually is. A robot is not a metal person: it is a metal non-person that looks like a person and is moved by external forces, like a marionette, to act in a fashion that we, human observers, symbolically associate with person-motion. The association is purely symbolic and exists in the minds of the observers only, not in any physical properties of the non-living thing being yanked through meaningless motions. Example: the gears in a clock do not care whether or not the clock is keeping accurate time–only to a human observer can a clock be “accurate” or “inaccurate” or “trustworthy” or “true”. Trueness is not a physical property but a symbolic one, since symbols define the relationship between a thought and the object that thought represents.
“If every human in the world were replaced by such robots, human civilisation would continue, although nobody would now be experiencing its delights.”
If every human in the world were replaced by a shovels and mops, with hats propped on the mopheads and coats tied around the shovel handles, civilization could also continue in exactly the same way. Mops could be propped up behind the wheels of motor vehicles, or leaned against the check-out counters of fashionable shops, placed on thrones or behind the desks of world leaders, and so on. We could break the mops and place them on battlefields, or prop other mops in jails and atop gallows; and likewise we could dress mops in wedding dresses, perhaps putting whisk brooms in diapers in the cradles in the nursery, and this would be exactly like the tragedy and joy of the human experience of civilization in every way possible — except, of course, that shovels and mops are not alive and do not move and do not think and do not suffer pain, any more than puppets, manikins, marionettes, steam locomotives, clockworks, or robots.
It seems to me, once again, that you entire “argument” boils down to an ambiguity, merely a trick of words, where you use a word or phrase (such as, in this case, “human civilization”) to mean what everyone else means by the exact opposite of that phrase (such as, in this case, a world inhabited by manikins).
Tell me: if a mop wearing a coat wiggles in the wind, and if at a distance in a bad light, I mistake it for a person, are mops people? What about a manikin in a coat? What about a Mister Lincoln robot made by Walt Disney? What about a clockwork doll? What about a really, really well made clockwork doll with eyes that blink and nostrils that expel and inhale air? What about a well made clockwork doll with a cunningly fashioned record player that makes what sounds to a human observer to be words and phrases in what sounds like a human voice? Can that doll write poetry?
If the answer to all these questions is “no” why then do you ask me to assume without question the huge assumption that an EVEN MORE WELL MADE DOLL, if the record player was really, really complicated, would be alive and have its own point of view? The assumption is that a material explanation of the location of material elements is what creates and defines consciousness: but this is the conclusion that this assumption is trying to prove, which makes the whole argument circular.
My answer is that the Tin Woodman, R. Daneel Olivaw, C3PO, Helen O’Loy, an emulation of Robert Sawyer’s brain, and so on and so forth are self-aware when and if from his own point of view, one of them states he is self-aware, and the statement is true. More importantly, he has a conscience when and if he is aware the nature and consequences of his acts.
The empirical approach to discovering whether or not a robot is alive is and always must fail for the same reason the argument of a solipsist must fail. Solipsism is the pretense that all knowledge is empirical, leading to the conclusion that no knowledge of the existence of another person is empirical, therefore no conclusion that other people are real as opposed to moving manikins is legitimate. But the pretense is false. The knowledge that other people are real is a moral conclusion, knowledge that comes to us through our conscience, and those sad individuals who treat other people as unreal as sociopaths, defective individuals, not philosophers adhering to a strict empirical standard.
Applying empiricism where empiricism does not apply, as in the question of solipsism, leads to a wrong conclusion.
I must request, once again, that you carefully distinguish between what I actually claim, and what your radical materialists may claim. I have not, indeed, said that my robot is conscious and has a point of view. I have merely said that it can produce poetry.
Since our discussion has grown confused, let me try to re-summarise it. You said, in your post on “Blindsight”:
I took this as a claim that poetry cannot be formalised, that is, it cannot be produced by a Turing machine, and in particular it cannot be produced by an electronic computer. If I misunderstood you and this was not your intended claim, please say so now.
Now let me re-build the robot, in case its original definition has got lost in the discussion: It consists of an ordinary human whose brain is scooped out. I then replace the brain with a fantastically powerful computer chip, made of Principlium-dosed Unobtainium; this chip is programmed to simulate precisely the quantum amnplitudes of the original brain, by means of the ordinary Neumann-computer operations of adding voltages. Then, it has an interface to the original body, such that whenever the simulation has a simulated voltage spike heading down a simulated nerve path in the simulated brain, it stops at the “edge” of the simulation and a real voltage spike is sent down the corresponding external nerve; and likewise in reverse for the incoming nerve signals.
Now I do not claim, at the moment, that this robot is conscious; I wish to let that pass for the time being. But I do claim that, if the original were so inclined, it will occasionally put hands to keyboard and produce poetry; and I believe this falsifies your original statement, that poetry cannot be formalised.
Now, was I mistaken in my original interpretation of your paragraph on poetry? Or do you claim that the robot as written will not in fact produce poetry? (Please note, whether it experiences the poetry is a separate question. I am only concerned here with the possibility of writing a machine to produce poetry, as we have built machines to produce good chess moves.) Or is there a third alternative I am ignoring?
I further say that this robot may be instructed in the same way we instruct a human, by speaking to it, and that it will indeed refuse some orders; if you wish to have it do something, you must produce an argument that would have convinced the original. And I hold that this blurs the sharp distinction you have made between “inputs” and reasoning; the inputs to this robot are, indeed, reasoning, although there is nothing in it to be convinced.
My apologies for my straw man fallacy: mea culpa.
My answer is here
“Deep Blue most certainly can play Chancellor Chess, or no-en-passant chess; you merely need to change the algorithm where it checks what is a legal move.”
That is exactly my point.
If *I* have to change the program, then the program is nothing different than if I wrote a book: the matter is marks on the page, the thought is the thoughts in my head.
Deep Blue does not think.
It has marks, in this case, electrons on a disk or in a circuit. The marks move according to rule I make. The intelligence you ascribe to the machine is nothing other than the machinesmith.
To ascribe human characteristics to nonliving object and tools is either a bad poetic metaphor, an error in semantics, or a neurosis of anthropomorphism.
Can we agree that Deep Blue has no power to change the rules of the game you say it is “playing”? Can we agree that Deep Blue is different fundamentally from, for example, a dog, who can indeed violate his house-training, pee and the carpet, and who does indeed look guilty, or at least anxious, when Master is angry for it? Can we agree that the dog operates under its own power. It is animate. Deep Blue is moves by external motions. it is inanimate.
Can we agree that “animate” and “inanimate” are opposites? Can we agree that “animate” is not a special set or sub-class or sub-set of “inanimate”? Can we agree that an “animate” living entity is not merely a form too complex and too dizzying for Turing to understand of an “inanimate” nonliving collection of bits of stuff that bump against each other blindly? That there is something radically, fundamentally and essentially different from a man who writes a book from a man who trains a dog? I do not have to coax or threaten my book to make it as I wish. I have to coax and threaten the dog to wish to wish what I wish. At no point do I open the dog’s skull and rewire the “piss in doors” clockworks.
These are differences of degree, not kind. Deep Blue and the dog alike have inputs that enable you to change their internal state; because the dog’s inputs are of a familiar kind, shouts and kicks, while the computer’s inputs are unfamiliar to your instincts, you consider them to be of entirely distinct types, but it is not so. Had Deep Blue been intended for use as an example in philosophical debates rather than to beat Kasparov at chess, it would no doubt have been provided with sound sensors (“ears”) and natural-language processors to allow it to change its algorithm in response to verbal descriptions of chess rules (“understanding”). But never mind Deep Blue; I mentioned it only in passing as a way-station to my robot. The robot, although it is not conscious, can be instructed in the same was as a dog, or indeed a man: By shouts and kicks, or by reasoned argument. Yet I think you would not on these grounds consider it animate, nor should you. Consequently, the argument that Deep Blue cannot be instructed by coaxing is not the real reason it is not animate, and you ought not to make that argument, but instead confine yourself to the essential point of whether or not it has consciousness.
Again: I wish to introduce, for purposes of making the discussion easier, a distinction between intelligence and consciousness; I wish to use intelligence to refer to apparently-from-the-outside purposeful behaviour, and consciousness to refer to apparently-from-the-inside purposeful behaviour. Even if you dislike the usage, can we not agree that it is a useful one for discussing the operations of machines that imitate human action?
Is it bad that I am annoyed when you claim I don’t exist?
I don’t believe I made such a claim. Would you like to quote where you think I did?
What ought I (and what do you) conclude about the ontological status of my brain? My mind?
Your brain exists and is in-principle predictable modulo whatever quantum effects there are; your mind exists and is likewise in-principle predictable; the brain is fundamental where the mind is not.
Is your leg muscle predictable? And if so does that mean your journey is predictable?
Sure, and ’tis a rare organ indeed that is more “fundamental” than the organism that uses it.
In principle, yes. As for fundamental, all I mean is that brains can exist without minds, but not the other way around.
God takes exception to your assertion. That’s an example of mind without physicality.
I see. So it’s not that I don’t exist, only that I’m non-essential. : )
Seems to me that your distinction between “apparently purposeful from the outside” and “apparently purposeful from the inside” is only useful once you’ve granted the notions of appearance and purpose.
I believe John’s argument is that appearance means appearance to someone and purpose means being purposed or intended by someone, and that a machine is incapable of being either of those someones.
I don’t think anyone is arguing whether a machine can imitate some human actions. The question is whether merely material interactions can produce appearance, or purpose, or any of the other kinds of thing we sum up under the term “thought”.
In other words, when Deep Blue “plays” an excellent game of chess, I do not consider Deep Blue to have intended its “actions” — even though those actions, taken in and of themselves and bracketed from anything beyond the chess game, may be indistinguishable from a human player’s actions — and I quite reasonably congratulate the programmer, not the machine. This is because the only way we could mistake Deep Blue for anything other than a very sophisticated tool is to ignore everything in the world except the record of the chess game.
“This is because the only way we could mistake Deep Blue for anything other than a very sophisticated tool is to ignore everything in the world except the record of the chess game.”
Well said, and I wish I had said this myself.
It neatly sums up the “creative ignorance” which lies in back of most of the materialist arguments. By “creative ignorance” I mean the deliberate ignoring as illegitimate any knowledge, wisdom, or belief which does not come through an approved channel.
This creative ignorance is not only useful, it is crucial to the proper study of physics, which ignores and must ignore various types of causation aside from mechanical causes: we only want to know from the science of ballistics where the parabolic flightpath of a cannon shot will land the ball–science neither knows nor can answer why or for what purpose the cannot shot flies.
This creative ignorance is not useful in a conversation with a solipsist, who pretends, even while he is talking to you as if you are a rational creature, that he has no “legitimate” knowledge that you are a rational creature.
As far “illegitimate” knowledge, the solipsist will admit (if he is honest) that for the purposes of everyday life (unimportant decisions like whether or not to get married, join the army, conspire against a tyrant, or go to heaven or hell) those quotidian decisions are premised upon the existence of other rational creatures, and the solipsist accept the illegitimate knowledge as true, either provisionally or because he has no other choice. This so-called illegitimate knowledge merely is not integrated into his philosophy: the solipsist does not know how to characterize it, or how to distinguish true from false.
Talking to a materialist is like talking to a solipsist, except that the rationality of which he has no “legitimate” knowledge is his own.
One can arrive at the conclusion that intelligence is a material substance that emerges from complex clockworks if and only if we take the approach in real life we take in the thought experiment of the Chinese Room: namely, the approach where we limit ourselves to and only to that information which can be passed through a mail slot on a bit of paper. In real life, I could pound on the door, and shout “Ni Hou!” and if Searle answers in English rather than Mandarin, I might find out whether he spoke Chinese himself, or was just printing marks, meaningless to him, from a rulebook written by a Chinaman not present.
The whole thought experiment assumes you find a watch in the desert without a watchmaker, or you come across a fully-stocked Chinese room with the rules for Chinese written out in English, but then comes to the conclusion that empiricism cannot deduce the presence of the watchmaker from the watch, or the presence of the Chinese translator from the Chinese Room.
It is what we can call the “peering through a mail slot” view of philosophy. It assumes we limit ourselves to empirical information only, and ignore the rest of philosophy, thought, experience and knowledge.
Again I refer you to the robot that perfectly simulates your brain in electronics. Now we needn’t ignore anything. Is it a rational creature, or not?
Answer this first for me, please, just so I know what you mean by the words “rational creature”:
1. If you found a book that described my every nuance of my every thought, conscious and subconscious, down to the finest level of detail, would the book be a rational creature? Would it be an intelligent creature?
2. Suppose you were talking to my wife the ventriloquist on the phone, and she so perfectly anticipated everything I might say in answer to any question you might ask, would the little image she has of me in her imagination she uses to formulate her answers be a rational creature? Would it be an intelligent creature?
3. Suppose my wife the ventriloquist wrote down on the back of a very very large set of three by five cards every possible question anyone might ever ask me, and wrote down the answers on the back, and she hired Searle to stand inside the Chinese Room and slide the cards out of the mail slot in response to whatever question was asked, would the Chinese room be a rational creature? Would it be an intelligent creature?
4. Suppose for the sake of efficiency, each of the three by five cards was made into a punch card, which a purely mechanical system of pins such as you find within a timeclock could “read”, so that anyone asking me a question were simply slip the card into the mail slot in the Chinese Room. At this point, my wife can save money, since we can fire Searle, and merely have the timeclock spit out the card with the answer written on the back. Would the timeclock be a rational creature? Would it be an intelligent creature?
In order:
1. Intelligent but not rational.
2. Both intelligent and rational, for this reason: That image is using human brain-circuits to run, and therefore is at least in principle capable of consciousness. In a case of true split personalities, the two personalities may be very different people, but they both possess consciousness and intelligence. The internal model of you is a very limited case of this. In effect, when your wife imagines your response to a question, a slightly different person than usual occupies her brain; and that person has self-awareness.
3: Searle, presumably, is rational; but the room as such isn’t. It is, however, intelligent, same as the book.
4: Intelligent but not rational.
Note again that I am using ‘intelligent’ in the limited sense I’ve defined elsewhere, and not in the usual English meaning; in effect it is a technical term, and like ‘force’ or ‘energy’ in physics, one should take care not to confuse it with the colloquial meaning.
“These are differences of degree, not kind. ”
Wait, so the difference between “animate” and “inanimate” that is, between “is alive” and “is not alive” is a difference of DEGREE?
Degrees of what, praytell? Give me an example of something that is halfway between being alive and being not alive?
As for the rest of it, you once again retreat into the clumsy metaphor of pretending the dog is machine who is not alive, and you use the same clumsy words “inputs” to describe actual inputs (computer programming) as opposed to commands.
A clock is a tool that reacts and does not act when inputs alter the clockworks, either their position or motion, such as by winding a mainspring.
A dog a is a beast who, fearing his master’s punishments, learns the difference between good and bad behavior with a blow or with a shout.
To call a command (a meaningful symbolic communication between one mind and another, even a simple mind like that of a beast) the same thing as an input (a meaningless bit of matter pushing another meaningless bit of matter in a passive tool which, not being alive or awake or aware, has no emotional reactions to the input, or even any awareness of the input, because it is not an entity, merely a collection of bits) is misleading, and leads you to the wrong conclusion.
Your “argument” consists of continually applying the same grossly and laughably inappropriate metaphor.
If I build a simple machine, such as a microphone connected to an oscilloscope, and I connect it to a lock on my door, and I contrive the machine to open the door when I say “Open Sesame!” the door will open not because my words have meaning nor because it is a real entity that really understands when the command means, and really reacts to the understanding of my intentional command by intentionally opening the door. Only an idiot could not tell the difference between this machine, and if I hired a parlor maid named Sesame, and asked her, through the door, to open it.
According to your clumsy and metaphorical approach, we are supposed to pretend we cannot tell the difference between Sesame the Parlor Maid and a voice-activated lock; and we are supposed to pretend that since we cannot tell the difference, the two must be one and the same, and therefore Sesame the Parlor Maid is not different in kind from the voice-activated lock.
Your “argument” consists of a metaphorical description of what a dog is, as if a dog is the same as a dog-shaped mannequin. The only thing the metaphor leaves out is the dog.
And what will you say when I have made a copy of the dog so good, that you genuinely cannot tell the difference? If its brain is electronic and contains no consciousness, and yet it responds as the dog would to a shout or a kick, where then is your difference between electronic input and a shout? I suggest that this is a mere prejudice on your part; even if we grant that consciousness is separate from matter, still the perception of the consciousness must at some point rest on matter and its response must return to matter. Whether this is by means of a nerve impulse arising from the vibration of an organic membrane, or a spike of voltage from the pressing of a key, ought not to matter.
You appeal to willed ignorance, by pointing out that there are easy experiments one can do, to tell a Chinese Room from a man; and you say that this invalidates the idea. But in turn you ignore that the Chinese Room can be modified so that you would not, in fact, be able to tell the difference by your experiment; and then the question would still stand. In short, you ignore the purpose of a thought experiment, which is to spur thought by separating out the essentials. If you were shown the classic trolley problem, you would not (I hope!) respond by trying to “think outside the box” and save both the five and the one; you would understand that the purpose of the question is to think about when it is appropriate to sacrifice one to save many, and would answer in accordance with that purpose. But with the Chinese Room, you feel free to ignore the purpose of the question and focus on trivialities of the purely hypothetical setup, as though this invalidated the inquiry.
“And what will you say when I have made a copy of the dog so good, that you genuinely cannot tell the difference?”
I will say that a counterfeit dollar bill, even one that fools the Treasury Department, is not a real a dollar bill.
If you can fool the robot dog into believing he is a dog, then you might have something. I can easily be fooled. I have yet to understand your argument as it related to the point of view of the dog. Since an unreal dog cannot have a point of view, it seems (to me, at least) to be an easy deduction that even if I, from my point of view, mistakenly thought the dog was real, the dog, if only he were real, would know better.
A voodoo magician might believe that by destroying the doll or imagine of a truth, you destroy the truth. I am not a voodoo magician. No matter how you rephrase the syllogism, merely because Lincoln looks like Lincoln, and Disney’s Lincoln Robot looks like Lincoln closely enough to fool a child, it does not follow that the real Lincoln is the same as a Disney Robot from his own point of view.
This argument rests on a fallacy.
I understand the point of the Chinese Room thought experiment perfectly well. I am pointing out that the thought experiment presupposes the limitations of empirical knowledge, and that those limitations are false-to-facts. Rewording the thought experiment to restate it (in this version, for example, I am not allowed to knock on the door) does not change the fact that, from the point of view of the man inside the room, he does not know Chinese. Only from the point of view of the man outside, communicating through the mail slot, a man who by the nature of the thought experiment is not allowed to have the vital information allowing him to tell true from false, only that outside has even the least possibility of being fooled.
I am not sure what you mean by trivialities. I am pointing out the assumption the thought experiment makes before it begins: namely, that all knowledge is empirical and that no knowledge is a rational deduction from first principles. The hypo assumes that the man inside the Chinese Room is beyond what knowledge can know, therefore beyond what true and false can say. This is clearly a false assumption even on its own terms.
You may, if you wish, object that questioning the assumptions under a hypothetical defeats the hypothetical. To question assumptions I grant you is “trivial” within and only within the artificial constraints of a conversation where we agree to discuss the question the hypothetical poses and not to question the assumptions on which it rests. Such question is trivial because it does not change the answer to the hypothetical question.
First, my robot is intended to demonstrate the formalisability of poetry, which (it seems to me) you denied a couple of posts back – I quoted it in another comment here. This discussion is getting rather messy, and it’s easy to miss a reply.
Second, you make at least three claims. One is that consciousness exists; I cannot well deny it, nor do I. The second is that we know this apart from any empirical information; this seems wrong to me, because our awareness of our awareness is itself empirie. When Descartes says “I think, therefore I am”, the logic is impeccable but the first clause is an observation. Third, that you know, in the same non-empirical way, that my robot is not aware – well, here I must object. We know that some assemblages of matter are conscious; how do you know that the robot is not one such? You have admitted that one can build an artifical brain in metal, which will then house a true consciousness; and you have agreed that predictability of the material housing is not a bar to such an artifical consciousness. Why then should not an electronic computer be conscious? Is it indeed so obvious that a human brain is the only possible housing for an awareness?
When Descartes says “I think, therefore I am”, the logic is impeccable
It’s also circular. “Cogito” assumes the “I” in the conclusion.
Modern science regards this not as empirical, since it cannot be demonstrated to others, but “subjective.”
Rolf, you say: “… consciousness exists; I cannot well deny it, nor do I. [John says] that we know this apart from any empirical information; this seems wrong to me, because our awareness of our awareness is itself empirie. When Descartes says ‘I think, therefore I am’, the logic is impeccable but the first clause is an observation.”
Observation of what? Certainly not of consciousness. Its effects, yes. Consciousness itself, no. Knowledge of those effects is empirical, but not the knowledge that consciousness is their cause. You need non-empirical knowledge to reach that conclusion. You need the strictly subjective knowledge of your own experience of consciousness. You need the rational knowledge of logic to determine that your experience of causing those effects is not unique.
Indeed, validity of all empirical knowledge must rest upon your faith that the truth is discernable to the human mind. Long before we gain and make use of empirical knowledge, we need a bedrock of knowledge that cannot be had through observation, measurement, and quantification. So your knowledge that human beings are conscious creatures is true, but it is not empirical — because you need subjective and rational knowledge that you must trust before you can make any valid use of empirical knowledge to draw conclusions.
So, what is the source of that fundamental non-empirical knowledge? Matter in motion? That begs the question. By definition, the empirical is the movement of matter (at least until it loses its form and physical causation break downs in the quantum foam). Therefore, whatever the source is, it is not physical, and so not empirical. Though not physical, it exists. We surely use it, for it makes those fundamental rules by which we gain empirical knowledge.
It is precisely because this rule-maker is non-physical, there will never be a poetry-writing machine. Machines operate physically, period. Because they entirely lack any non-physical capacities, they are strictly rule-followers, not rule-makers. A machine that allegedly makes rules, whether in a thought experiment or in the real world, is actually a machine that follows a rule for making rules. And you can see that this leads to an unresolvable regression of rule-following if we insist that rule-making is reducible to the physical. The buck stops nowhere.
But there is more basic reason why there will never be a poetry-writing machine. Poetry cannot be written entirely by a set of rules, no matter how cleverly drafted. The best you’ll get is doggerel. This is because poetry is an aspect of beauty, and beauty is a quality, eluding empirical reduction to enumerations, measurements, and quantifications. Beauty is brought about and discerned by mind alone. Yes, we do see beauty in the physical, such as a sunset. But that is only because the physical is properly ordered by the rules it must obey — and as argued above, that rule-maker to which matter must submit cannot be more matter. The rule-maker is a mind.
Regards,
Bill T
The argument that “I think therefore I am” is an “observation” cannot be true until and unless we define the awareness of thought or self awareness as an empirical observations arrived through the senses. In order to do that, we have to define the word “senses” to mean exactly and precisely that type of awareness which does NOT come into us through our senses, namely, thought, apperception, introspection, imagination, conscience, reason.
This is awkward, because then we are defining the word “senses” (sight, hearing, touch, smell, taste) to include those things specifically excluded by the word (thought, self-awareness, reason, imagination, conscience). As with all misleading changes in vocabulary, changing the word “senses” to mean both sense-impressions and non-sense-impressions requires that we then define new words, such as “internal senses” and “external senses” to re-establish the difference that the new misleading change is trying to blur.
In any case, whether Descartes is in a house or with a mouse, in a box or with a fox, on a train or in the rain, in the dark or up a tree or with a goat or on a boat, no matter whether Descartes is here or there or anywhere, the statement “I think, therefore I am” is still true. All other statements concerning sense impressions (or, if we insist on a misleading terminology, all other statements concerning “external sense impressions”) can be changed without changing that statement of self-awareness (“internal sense impression”). If A and B and C and D all lead to X, then X is not dependent or A or B or C or D.
Not to get too technical, but even those categories of perception which require some content, such as our category of extension or space (up, down, left, right, to, fro) are independent of any specific sense impression. It does not matter whether we see a mouse or a house, a goat or a boat, the mere act of seeing anything imparts the information that all things do not occupy the same space, which in turn implies that the category of “space” is necessary for sense perception.
The “cogito” of Descartes likewise does not care what you are thinking: the mere fact THAT you are thinking demonstrates beyond any logical refutation you exist, because non-existing things cannot think.
There are technical terms in Kant to describe these precise relations, but I will not trouble the reader with them at this time.
The argument that Descartes’ “Cogito Ergo Sum” is an empirical observation is a misstatement based on sloppy vocabulary.
It would be like a physicist who said that everything in the universe was made of neutronium on the grounds that all waves are atoms, and all atoms contain neutrons: well, I suppose if we agree to label every type of subatomic particle a “neutron” we can defend this goofy statement, but no real physicist would tolerate a discussion with such slovenly terminology.
In reality, only neutrons are neutrons, and other particles, such as electrons and protons, are something else.
Likewise, in philosophy we have a technical vocabulary designed to avoid sloppy reasoning based on ambiguities, puns, and plays on words: in philosophy, we do not use the word “empirical” to mean everything both empirical and non-empirical, and then make the goofy argument that Euclid’s deduction (ELEMENTS I. 15) that vertical angles are equal are based on an “observation” of the abstract and formal properties of a triangle using the “sense impression” of the imagination and the reason.
Empirical observations are contingent (depend on the sense impression, and change when the sense impression changes) whereas rational deductions are necessary (depend on their axioms and common notions, and do not change).
Let us separate what I said from what I did not say:
1. Consciousness exists. We agree on this point.
2. We know this apart from any empirical information. This follows directly from the definition of “empirical.” I neither see, feel, smell, nor taste my thoughts. I think my thoughts. Thinking is not a sense impression, indeed, the human race as invented a word ‘sense’ for the sole purpose of distinguishing between things we see (external objects) and things we think (symbols).
3. In your hypothetical, you hypothesized only that there is a doll that looks to me like a dog (i.e. when I look at it, I make the artificial and arbitrary symbolic association with the idea “dog” and that I misapply it to the dog-shaped doll) and that I am dumb enough or ignorant enough to mistake or mis-categorize it for a dog because a cunning artificer has equipped with springs to make the legs move, or phonograph records to repeat a barking noise, and so on. In other words, so far in the argument, you have hypothesized a doll that no one who looks closely would mistake for a dog, and then you ask me whether the doll is a dog. Well, maybe I misunderstood your hypothesis, but you hypothesized a doll that was not a dog.
You can, if you wish (it is your thought experiment) hypothesize that the doll is self-aware. Likewise, I can hypothesize that rocks are self aware and obey Newton’s Third Law of Motion of their own free will, because they fear terrible punishments from the fairy-king Oberon, who sees any rock disobeying. But such an hypothesis would not be scientific reasoning, which disregards final cause by definition.
You ask whether an exact enough replica of my brain would be self-aware. You tell me what self-awareness is, and I will tell you if an exact replica of my brain would have it.
You can use the word “intelligence” to refer to the intelligent behavior of tools being used by creatures with intelligence, yes.
It is, however, a misleading use, because each time you slyly try to slide in the implication that the tool, not the tool-user, is “intelligent”, so that you can draw the absurd conclusion that the “intelligence” in the tool-user is the same as the “intelligence” in the tool, I raise the same objection, over and over again.
It is also a use not useful for any discussion aside from one like this, where you are trying to factor consciousness out of the equation by using a passive-voice metaphor which pretends consciousness does not exist.
We live in a world where consciousness can produce intelligent actions in men, and where tools made by men can behave in a fashion you call “intelligent”. That is to say, where A stands for consciousness and B stands for tools and C stands for intelligent behavior, we live in a universe were A implies C (as when a man uses his fist to hammer something) or A implies B implies C (as when a man builds a hammer to hold in his fist to hammer something).
Your reasoning, if I understand it, is that since B implies C (the hammer intelligently acts to hammer nails — the nail-hammering is intelligent behavior, Deep Blue plays chess) and that since A implies C (A man can hammer with his fist; Kasparov plays chess) therefore A is B (what at first appears to be conscious and deliberate actions on the part of a living man is in fact no different than the reactions of mannequins or clockwork automata).
If that is your reasoning (and I hope it is not) it suffers from an obvious formal defect (undistributed middle) which become obvious once we cure the informal defect (ambiguity).
Honesty will allow you or anyone, of course, to coin a term which combines as if they were one thing something that in nature is two things, provided only that we keep in mind, and warn our readers, that the distinction between the two things still exists.
There comes a point of complexity where you are no longer able to keep all the variables in your head, and then, when asked why the program did X instead of Y, you would be unable to answer. And I further suggest that all useful programs are like this to some extent; because if you knew what the output would be, why bother to run the program?
First, If you can pound a nail into a board with a rock, why bother to use a hammer? A computer does mathematical calculations far more quickly and efficiently than a human person, just as a hammer pounds nails far more quickly and efficiently than a rock (to say nothing of a bare hand).
Next, complexity is not a magic wand that can save AI. It doesn’t take very long for me to be able to lose track of variables and their outputs. That’s why I write stuff down. That’s why I make mistakes when I’m programming a computer. That’s why I print out the lines of code and run stack traces and so on. All of these are tools that allow me to deal with massive complexity.
Complexity does not, in and of itself, make intelligence. It makes complexity. A rubber band ball can be very complex, in that it has many layers and many interactions between its component parts, but that doesn’t make it anything more than a rubber band ball. Likewise, a computer or computer program can be massively complex insofar as it has many layers of functionality and many interactions between a wide variety of inputs and its functions; but in the end it remains a tool, and it acts according to the laws of nature and the rules of its programmer. It is the programmer, the toolmaker, who put together the steps of handling inputs and pushing them through to outputs, and it is the programmer’s thought that we see in the machine’s actions — even if he doesn’t know concretely what the outputs will be for any given input.
Does this mean that mechanical AI is impossible? Not necessarily; but “complexity” is not a sufficient cause. Likewise, just because the brain has some predictable neuro-electrical events does not mean that those events are sufficient to explain thought.
“What I claim, however, is that all the external behaviour of a human can be formalised in this manner. ”
Since this is the point I have been stressing from the beginning of the conversation, I am frankly puzzled that you speak of it now as if you are introducing it, and frankly baffled that you think this point helps rather than defeats your argument.
Are you arguing that a fool cannot tell the difference between a mannequin and a man? Are you arguing that if a programmer programs a moving mannequin to make a moving speech, such as, for example, the ‘Meet Mister Lincoln’ exhibit at the World’s Fair, the audience might not be moved?
So if your robot carries out an algorithm to react to chess moves with a move that a chessmaster invented, or carries out an algorithm to write down doggerel poetry according to a mechanical system of rhymes a poet invented, or squirts water from the round painted orbs the mannequin-maker intends onlookers to pretend are symbols to represent eyeballs, what follows? Does it follow from this that human beings are dead mannequins? Does it follow that strict empiricism cannot distinguish the difference between human beings and dead mannequins? If so, so much the worse for strict empiricism.
No, build me a mannequin that can suffer sorrow from its own point of view, and then describe its every action, including its purposeful and intentional behaviors, as if they are mere reactions to external forces, and include nothing of purpose nor intention in your description, and you will have made your case.
If the radical materialist argument is that material objects, such as pictures are mannequins or robots, can be used by living souls to symbolize or represent or go through the motions of what living souls do, the argument is trivial.
Well, now I must admit to confusion as to what you are really claiming. In your post on Blindsight, you said this:
This seems to me to be a claim that poetry-writing cannot be formalised. Did you, perhaps, intend it only as a claim that the experience of appreciating poetry cannot be formalised? If not, well then I say that the act of putting pencil to paper and writing down a (fresh, new-insight-containing, perhaps even good) poem is an external behaviour of human beings, and formalisable by means of my quark-level robot; and what’s more, that if it can be done by such a robot, then it can also be done by a simpler algorithm, because the robot is intended to simulate all the possible behaviour of a human, and poetry-writing, like chess-playing, is only a very narrow subset of that.
You are impatient, you run ahead, you attribute to me claims that I have not made, because you think I ought to make such claims. Let us first settle this question of poetry, and then see what else we might conclude.
Again, perhaps we are agreeing violently rather than disagreeing. As a side note, our conversation now has many branches, and no doubt it is easy for either of us to miss something the other says; possibly it is time for a new summarising post?
Maybe we can call this the I/O fallacy.
I’ve considered this issue whenever I hear people talking about how studies and injuries to the brain prove that there is no such thing as a soul. Balderdash.
Suppose for a moment that you were conversing with someone through an instant message program. During this conversation, you notice that the other person never types the letter ‘r’. Why? Is it that the person has some condition that keeps them from recognizing that letter? Or could it be that they have a damaged keyboard where there is no letter ‘r’ present at all?
In this analogy, the world is the IM program, the computers are our bodies & brains, and the people typing on them would be our soul. At best we can only tell the input/output of the program and the devices used to interface with it. Just because we can see that damaging a keyboard causes later input to be distorted doesn’t prove whether there is anything behind said keyboard or not.
You know, that might be a nifty idea. . . .
More to the point, Deep Blue did not defeat Kasparov. It did not really pass any kind of Turing-like test, and there’s also a great likelyhood that the IBM-team (IBM funded the entire event and used it as PR) cheated.
Kasparov had defeat the *original* Deep Blue a year before. When they rematched, it was composed of not one, but two server blade towers, along with a much larger staff. Which included two chess grandmasters.
Do you see where this is going? There is also the fact that the staff did almost everything behind the scenes, behind closed doors where nothing could be observed. And the fact that, suddenly in the third match, after Kasparov had soundly trounced the machine twice in a row, it suddenly began displaying very humanlike intuition and cleverness in it’s moves. It is also no accident that, after acussing IBM of cheating, they promptly dismantled Deep Blue and shipped it’s halves away to company storage sites, where they are now museum pieces.
Kasparov himself, the greatest chess mind ever to have lived (sorry, Fischer fans), concluded that the IBM team had in fact began intervening on behalf of the machine, helping it. It is no accident that one of the grandmasters on the team just happened to have been defeated by Kasparov the previous year.
The entire affair was designed solely to raise IBM’s stock prices. Nothing was proven, there was no great moment in history where machines were finally more intelligent than humans. It was a sideshow designed to amuse and to make money, at the expensive of the greatest chess mind ever known.
Kasparov knew exactly how he’d been screwed, and he never really recovered from the insult to his ego that IBM and the subsequent press dealt him.
Fritz Leiber wrote a story about that many years before it happened, called “The 64 Square Madhouse”
You wrote: Also, there is a formal error in logic here, called the fallacy of the undistributed middle. The syllogism “(1) Men’s thoughts can be described in terms of complex decisions or formalization (2) Computer program logic can be described in terms of complex decisions or or formalization therefore (3) Men’s thought are computer program logic.” suffers the same error as the syllogism, “(1) Englishmen are men (2) Frenchmen are men therefore (3) Englishmen are Frenchmen.”
I think that this contains a category error, namely “algorithms” are a fundamentally different thing from “men.” Let me offer some, admittedly half-baked, ideas on this. I am a software engineer by training, profession, and desire. To fill a gap in my education, I recently designed my first digital circuit (an adder) from scratch out of NAND gates. My daughter saw my circuit diagram and asked what it was. I told her that it was a representation of a digital circuit. Since she just started college and is studying graphic design, and because Hofstadter’s GEB was on my mind, I showed her Magritte’s painting “This is not a pipe.” But that got me thinking. I had five different descriptions of a digital adder: a circuit diagram of NAND gates, a LISP expression from which the circuit diagram was drawn, the boolean expressions for the summation and carry portions of the adder in NAND form and NOT/AND form, and the truth tables from which the boolean expressions were constructed. All of these were descriptions of the circuit — but they were not the circuit.
But this led to the question: are there things for which the description of the thing is the thing itself (where ‘description of’ is loosely defined)? I came up with two things that, I think, fit the equation. The first is God. Certainly the Trinitarian view is that “what God says” is identical to “what God is” {“and the Word was God”}. I think the second is intelligence. The Turing test is based on this equality, even if this equality isn’t explicitly stated in the test. This connection surprised me at first. I didn’t expect it but, upon reflection, it makes sense.
In both of these cases, the “image” of the thing is identical to the thing and this is why I think your particular objection misses the mark.
If someone had asked me ten years ago if we would be able to develop a human-like artificial intelligence, I would have said “no” because I thought that intelligence was “composed” of mysterious “God stuff” that we would not be able to duplicate. Today, I lean the other way. I think that we will discover that randomness underlies far more of human “intelligence” than we like to admit (cf. God, The Universe, Dice, and Man) and that we will find an algorithmic basis for morality (cf. Artifical Intelligence, Evolution, Theodicy). These thoughts are still in embryonic form, but I think they have tremendous potential.
Given advances in vision technology, the density of computer memory, processing power, and algorithm development, I think that one day human level artificial intelligence will be real. We will have created man in our own image.
“Given advances in vision technology, the density of computer memory, processing power, and algorithm development, I think that one day human level artificial intelligence will be real. We will have created man in our own image.”
Not to trivialize the accomplishment of such a feat, but every nubile woman already has this power in her womb.
The difference between putting together a baby one electronic circuit at a time and putting together an baby using sperm and egg with biological material does not strike me as significant. And (I hate to say it) not something that has any bearing on the philosophical discussion of the difference between mind and matter, meaningful and meaningless, rational and empirical, alive and dead, things that act of their own nature and things whose nature is to react.
The point that strikes me as significant is that we, a crassly materialist and philosophically retarded civilization (we are far more crude, theologically than the Thirteenth Century — see the Holocaust for details), would tend to treat such children as if they were tools or machines, when, like the baby in the womb, at the point at which they became self-aware, they would, for all practical purposes, be alive.
Not to blow my own trumpet, but my meditations on this matter are on display in my book THE GOLDEN AGE, where I assumed that any civilization that treated its artificial people as mere tools could not compete and would be overshadowed and absorbed by any civilization that was more civilized about how it treated its children. (Especially if the children were more intelligent than the parents, and, in the long run, far more powerful).
I don’t disagree with anything you said; however, you didn’t address my main point, namely, the reasons for why I think your argument about the undistributed middle being wrong. For intelligence, and the Trinitarian view of God, the image of the thing is the thing.
A convincing argument if I believed in your god.
Again, convincing if I believed in a creator – and perhaps not even then. Not every theist believes that their god guided evolution; it follows that they believe, with atheists, that unguided natural selection can produce purposeful-from-the-outside behaviour.
A convincing argument if I believed in your god.
The point is that your “minds cannot exist without brains” is itself an unconvincing argument.
Not every theist believes that their god guided evolution; it follows that they believe, with atheists, that unguided natural selection can produce purposeful-from-the-outside behaviour.
So? Truth has never been determined by numbers. In any case, there can be no purpose without consciousness. “Purpose” is nothing more than the determination of a particular path through some state space by a self-aware articulate goal seeking entity. Therefore, in order to have “purposeful-from-the-outside” behavior, what you call “consciousness” has to be somewhere in the system. So you’re claiming that a purposeless process (evolution) produced purpose. Conservation of information theorems make that an extremely dubious proposition.
I did not say that as an argument, but as a conclusion; I was asked what I believe about brains and minds, and answered. If I had been asked why I believed so, I would have given an argument, chiefly that I’ve never seen nor heard convincing accounts of a mind without a brain.
Nor did I say so; I was merely pointing out that your argument doesn’t necessarily even convince someone who shares your axioms.
The usual weakness of the English language. If there were no humans, an anthill would nonetheless act in such a way as to bring food into the hive, waste products out, and eventually to produce more ant queens to continue the process. It’s fine to say that there’s no purpose in it, but we do need a word to describe what’s going on; I suggest ‘intelligence’.
Here you are mistaken, because you haven’t considered the effects of filters. It is unfortunately rather a common habit in theists to approach information theory, entropy, and thermodynamics without any mathematical grounding; I request that you not do so. Evolution can produce intelligence in the sense mentioned above by killing off anthills (and primates) that act in apparently-purposeless ways, or unintelligent ways if you prefer. I argue that it can also produce consciousness as a byproduct of such intelligence; note that the consciousness doesn’t add any information to the system. But that is all irrelevant to my original point about the robot copy, which was intended only to demonstrate the formalisability of human behaviour, including poetry.
Nor did I say so; I was merely pointing out that your argument doesn’t necessarily even convince someone who shares your axioms.You were pointing out a number of people who don’t agree with an argument to show that it isn’t necessarily convincing, but this isn’t an appeal to numbers. Right.
The usual weakness of the English language. If there were no humans, an anthill would nonetheless act in such a way as to bring food into the hive, waste products out, and eventually to produce more ant queens to continue the process. It’s fine to say that there’s no purpose in it, but we do need a word to describe what’s going on; I suggest ‘intelligence’.
I understand that. But given the connotations of the word “intelligence” in the English language, it looks like you’re trying to sneak in the back door what you want to keep out of the front door. Ants are machines. Without (self) purpose or intelligence (at least in the Turing sense).
Here you are mistaken, because you haven’t considered the effects of filters. It is unfortunately rather a common habit in theists to approach information theory, entropy, and thermodynamics without any mathematical grounding…
So much for my engineering degree in math. Unfortunately, it’s far too late to demand a refund.
<…note that the consciousness doesn’t add any information to the system.
Of course it does. Your definition of consciousness includes self-awareness. Where did the information of “self” come from?
… which was intended only to demonstrate the formalisability of human behaviour
A point echoed by my reply to John.
But the anthill does exhibit purpose. The ants act do nopt gather food to no end. (The anthill, however, is just a pile of dirt.) Purpose or telos need not be self-conscious, or even aware. Even inaminate objects evidence “towardness”.
Don’t forget “instinct” as an alternative to “intelligence” but don’t fall into the trap of Descartes machine-image “instinct.”
unguided natural selection can produce purposeful-from-the-outside behaviour.
The reasoning is circular. How do you know natural selection is “unguided”? It tends always toward an end; viz., better reproductive fitness for a niche. When A leads to B consistently and not to C or D or nothing at all, there is towardness in nature. The deduction is that nothing is orderly by chance. As Einstein once wrote to M. Solvine, a priori we would expect the natural world to be chaotic. That it is deeply ordered is what he called a “miracle.”
Minor linguistic point: I was using ‘anthill’ to refer to the assembly of all the ants in a hive, much as I might refer to “America” doing something without any expectation that the actual continent would rise up.
As for ‘instinct’, it doesn’t feel natural to me in this context; rather it feels like an invention of the Victorians to conceal their unease at mere animals exhibiting intelligent behaviour.
By noting its many failures. First, observe that it is a filter, not an engineering approach. If I throw a hundred dice and pick out the sixes, that is different from making loaded dice that always get a six, even though the end result is twenty-or-so sixes on the table. Second, by noting that it never looks ahead; it is a locally greedy algorithm. If protein A would preserve a penguin through the Arctic night at half the energy cost of its current approach, but depends on protein B to work; and protein B has the unfortunate side effect of making the eggshells slightly thinner – then protein A never evolves. If larger incisors make it easier to kill elephants by slashing, and thicker hide makes it harder, then you get larger teeth and thicker hide until the cats all die for lack of calcium to build the teeth. If protein X makes male rats larger and more aggressive, so that they always win in mating fights, but has the side effect of making all the offspring male, then the genes for protein X will nonetheless spread like wildfire through the rat population until there are no more females, and the area is recolonised from the outside. And third, observe that it doesn’t produce “good”, but only “good enough”. Like the man in the joke who doesn’t need to run faster than the bear, only faster than his brother, it will never produce anything one iota better than the absolute minimum that can get the job done – whatever the job happens to be. Thus we get such spectacles as the human eye with the light-sensitive nerves on the wrong end; the human spine designed as a suspension bridge when it needs to be a supporting pillar; and of course aging at age forty because there’s no particular pressure for the organs to last longer than 30 years or so, enough to produce children.
If this is guidance, I hate to think what chaos would look like.
Nah. It’s an Age-of-Reason, Scientific Revolution distortion. Aristotelian instinct was much more than the clockwork animates of the Enlightened.
+ + +
By noting its many failures.
How can a natural law fail? Can gravity fail? What about electromagnetism?
it is a filter, not an engineering approach. If I throw a hundred dice and pick out the sixes, that is different from making loaded dice that always get a six
An engineer cannot design a filter into a system? How is this a failure?
Second, by noting that it never looks ahead
Strangely, by never looking ahead it still manages to improve reproductive fitness. How is this a failure?
And third, observe that it doesn’t produce “good”, but only “good enough”. It will never
produce anything one iota better than the absolute minimum that can get the job done
But why should it do more than that? How is this a failure?
Thus we get such spectacles as the human eye with the light-sensitive nerves on the wrong end
What makes it the “wrong” end?
If this is guidance, I hate to think what chaos would look like.
Yes, you certainly would. A world where the light sensitive nerves do not hook up at all or even develop as light sensitive; where heavy bodies could fall up as easily as sideways or down; where a human “spine” never becomes a spine at all, but bits and pieces of bone distributed randomly throughout a mass of growing tissue.
BTW, what do you suppose “guidance” ought to look like? Everything comes out the way you think it should? When you work though an instrument you impose on yourself the limitations of that instrument. Geometers could not trisect the angle because they had limited themselves to compass and straightedge. With protractors and rulers, it is simple. But there is greater art in achievement within boundaries. If you work through genetics, say, the spine of the upright ape will be developed from the spine of a quadruped. It doesn’t poof into being from nowhere. That would not be guidance, but poofing.
Failures to produce the sort of good results that I would expect from an engineered, as opposed to a natural, process. In this sense, gravity does indeed fail to produce anything except boring, least-potential-energy spheres.
There are many cases of natural selection entirely failing to improve fitness, where a procedure that looked ahead would have done so. Shall we count that also as success?
There is light-absorbing (non-transparent) tissue between the light-sensitive bits and the outside, thus making the eyes less sensitive than they might be.
This appears to refer to a different kind of guidance entirely. Please try not to confuse different concepts in one conversation. Also note that the kind of horror you describe does indeed happen; we refer to them as mutations, and they usually die while still in the womb.
Failures to produce the sort of good results that I would expect from an engineered, as opposed to a natural, process.
But again, you are assuming a priori that “natural” is not “guided.” I’m not sure why you think it needs to be “engineered” or why you think you could have done better. Give it a shot, why not? Remember that you have bound yourself to work with materials as they are, not as you wish they might be, and that new forms are to be derived from prior forms.
In this sense, gravity does indeed fail to produce anything except boring, least-potential-energy spheres.
So natural laws fail if they are boring?
There are many cases of natural selection entirely failing to improve fitness,
where a procedure that looked ahead would have done so. Shall we count that also
as success?
Holy moly! You have falsified the Darwinian theory!
(Yes, you count it as a success if the organism succeeds in reproducing and/or attaining its natural ends. There is no requirement that they do so in a non-boring way, nor even in what you suppose might be a more “efficient” way.
There is light-absorbing (non-transparent) tissue between the light-sensitive bits and the outside, thus making the eyes less sensitive than they might be.
What would happen if they were more sensitive to light? Actual design engineers know there are often trade-offs. When there are multiple Y’s to optimize, it is not always possible to target the X’s to achieve simultaneous optimization of all the Y’s.
This appears to refer to a different kind of guidance entirely. Please try not to confuse different concepts in one conversation.
You said that “if this [results that you think you could have done better] is guidance, I’d hate to see chaos.” Now you say there are “different kinds of guidance” and chide me for apparently using a kind other than the one you want me to use. What are these “different kinds of guidance”?
We seem to be dealing with at least three possible forms of guidance here, of which you have introduced two. Let me enumerate them:
1. Guidance in the sense of forward-looking engineering; in the case of my imaginary penguin proteins, an engineer would be willing to trade off some eggshell thickness to fix protein B now, in exchange for the future efficiency advantage of protein A. Evolution does not do this. This is the form of guidance of which I deny the existence.
2. Guidance to mean “The consistent operation of natural law”, as opposed by your example “[A world] where heavy bodies could fall up as easily as sideways or down”.
3. Guidance meaning “Biology works as designed”, as opposed by your example of “a human “spine” never becomes a spine at all, but bits and pieces of bone distributed randomly throughout a mass of growing tissue.”
To refer to the second two as ‘guidance’ at all is, it seems to me, a theistic-ish thing to do; an atheist doesn’t consider them to be guidance, but just the way things are, with nothing intelligent or conscious. I believe there is a technical term for this idea in theology, but I can’t bring it to mind at the moment. However, it is not something that can be settled by any evidence. The first kind of guidance is, however; and that is what I was referring to. I hope you will not confuse matters further by attempting to bring in these other things and call them guiding evolution; equivocation is the term, I believe, and it is a fallacy.
I phrased myself badly. I should have said that there are cases where a species has evolved itself to extinction, as in the case of the Y-driven segregration distorters I mentioned in rats. This is a special case of the more general rule that genes ‘do’ what is good for them in the short term, not what’s good for the species they inhabit in the long term, much like viruses in a host.
an atheist doesn’t consider them to be guidance, but just the way things are, with nothing intelligent or conscious.
That’s why we can explain music without regard to the “musician hypothesis.” Compression waves in the air is “just the way things are.” They can be described in purely natural terms. There is nothing “intelligent or conscious” about the reed or the string. Just molecules in motion.
In the same manner you can explain whittling by pointing to the knife. Everything experienced by the whittled object can be explained by aspects of the knife — angle of the blade, sharpness, speed, direction, and depth of cut, and so on — without need for the whittler. There is nothing “intelligent or conscious” about the knife.
The same strategy was tried with “altruism.” It was defined in such a manner that it was defined out of existence, just so some folks committed against it could say there was no such thing as “real” altruism.
But if an agent guides things through an instrument — whether it is a whittling knife, a clarinet, or nature — it should be obvious that when things work toward an end it must be the result of an intelligence, and there is no intelligence in nature, whittling knives, or clarinets.
+ + +
I believe there is a technical term for this idea in theology, but I can’t bring it to mind at
the moment. However, it is not something that can be settled by any evidence.
You may be thinking of teleology. But it is a metaphysics term, not theological, as such. Aristotle, who had no theological commitments, thought it pretty damn obvious that there was telos in nature and that the evidence was abundant.
Without telos, there can be no efficient causes, no natural laws, only chaos. We cannot say that A is a cause of B unless A is directed toward B in some fashion. Otherwise, as Einstein wrote to Solvine, one would expect chaos, not order, in nature.
+ + +
I should have said that there are cases where a species has evolved itself to extinction
How is this a failure of evolution? Is not natural selection supposed to weed out the less fit? Evolution via natural selection is supposed to be a global cause. Don’t look for some locally-defined “success.” The end of evolution is greater fitness, not any particular species. In the course of this the lesser fit are transformed or die off.
genes ‘do’ what is good for them in the short term, not what’s good for the species they inhabit in the long term, much like viruses in a host.
Genes don’t “do” anything. They don’t even exist in the quasi-atomistic 19th century model. They are sequences of nucleotides on DNA strands. They do not act in isolation, but as a whole. They are used by an organism in an environment. Only then do they produce their proteins. Two populations with identical genomes – clones – can express themselves differently depending on environmental conditions. This idea of genes as independent replicating units is one of the more useless notion Dawkins came up with. (Perhaps only ‘memes’ is more useless.)
But the anthill does exhibit purpose. The ants act do nopt gather food to no end. … Purpose or telos need not be self-conscious, or even aware.
If English means anything, it does need to be self-conscious. The only “purpose” exhibited by the ant colony is to the self-aware observer.
What Shannon or Kolmogorov information is needed to describe it?
You will pardon me if I thought I made this point better.
What Shannon or Kolmogorov information is needed to describe it?
I don’t yet know. I haven’t finished creating the algorithms for a human level intelligence. I may never finish. Nevertheless, you said, … intended only to demonstrate the formalisability of human behaviour. If it can be formalized, it can be expressed in bits. Maybe you should answer your own question?
Well, can you suggest a better word, then? I would hope that, for a purely technical discussion, we might be able to ignore the usual connotations of a word and adopt a definition not quite the same as the usual one.
We might be able to ignore the usual connotations, but we usually don’t. Old habits die hard. Ants are turing machines with limited programming.
Well then, you should be able to consider how the filter of natural selection works, no?
I know what’s claimed for it.
You surely did not intend to say that evolution is forbidden by thermodynamics
No. The earth isn’t a closed system.
What prevents the same argument being made for purpose?
Not a thing. Whether or not it’s a valid argument is another matter. It typically ends up being an argument about probabilities.
Indeed, I will try to do so. Consider that the robot is, if anything, more complicated than the original human, since there must be some description of the computer/flesh interface; yet it is not conscious. It follows then that consciousness does not require any additional information at all; you get it automatically from a particular kind of complex structure where you only need to describe the atoms. (Which is not, of course, to say that all descriptions of the atoms would have the consciousness.) Epiphenomena are like that, you get them for free without any additional information.
Consider that the robot is, if anything, more complicated than the original human, since there must be some description of the computer/flesh interface;
Surely that’s an irrelevant complication. It’s the software, not the particular form of the hardware, that makes us human.
yet it is not conscious. It follows then that consciousness does not require any additional information at all; you get it automatically from a particular kind of complex structure where you only need to describe the atoms. (Which is not, of course, to say that all descriptions of the atoms would have the consciousness.) Epiphenomena are like that, you get them for free without any additional information.I must disagree, since it’s the connections between the parts that control what you call the epiphenomena. F(N X IC) is a description of “atoms” that in a certain context is meaningless. (INCF X) are the same atoms but it performs a useful function. Consciousness requires that the wires between the NAND gates be arranged just so. That arrangement is the information that has to go into the system.
Well yes, but the arrangement is what I meant by “description of the atoms”. I did not mean to say something like “Carbon atoms have 6 protons and 6 neutrons” and so on, but rather “There is a carbon atom at position X, and hydrogen at X’, and their excitation states are this…”
It’s clear that you are being unfair to the “radical” materialists.
(And, in case that was too cryptic)
wrf3: You’ve built your argument on the claim that the representation of the thing is not the thing, but I’ve given you two counter-examples, which you haven’t yet addressed.
I shall step up to the plate.
I deny that meaningless numbers and symbols can produce meaning. Numbers and symbols do not and cannot have meaning without context, and that context must not consider of other meaningless numbers and symbols.
Take the string “1111111″. It could refer to the number 255. Or it could have been counting the quantity of an item, having begun from zero, and actually means that there are 256 items. Or it could be a flag byte of a berserker packet. Or it could be the 6502 instruction BRK. Or it could be the decimal number 11,111,111. Or it could be a piece of an encrypted or compressed message. All of these are equally possible. We simply have no clue.
Take the string “#include ” It is, almost assuredly, an include directive. Almost. It could be a confusing python comment, or be turned into something else entirely via a macro trick, or a function declaration in some hideous esoteric language. While some may be more or less likely, all of these are possible.
Let us even suppose we had the object code of the program reading our string of “#include “. If we knew what processor it was for, we could, once and for all, determine what it means. But only if we knew which processor. There are an uncountably infinite number of possible processors, such as “x86 except the twenty-seventh interrupt is ignored.” One or more may be more likely, but mere likelihood is not the same as a universal meaning. The evil space socialists from Cygnus may have their own desires in CPU design, such as irritating philosophers, or maybe they don’t like being interrupted for the twenty-seventh time.
Let us even suppose we have a perfect map of the physical embodiment of the CPU. Have we now a solution? No. For though we can say “The CPU will progress from electron state A to electron state B in so many nanoseconds.” we know not what either electron state A or electron state B mean without reference to the hardware.
So we move on to videocard, then the monitor, then the photons flying into the eye, and then through the nerves into the brain. Now can we determine the meaning? If we are limited to the material world, no. For we can say “Brain Area 3379 is now active.”, but what it means when “Brain Area 3379″ is active we cannot say. We may discover that certain facial muscles move in a certain way when Brain Area 3379, but then we would need to discover what the facial signals mean. And to determine that, we would need to study other brains in the area, and in each of them we would face the same situation!
No matter how many meaningless symbols we are provided with, we cannot draw meaning out of them any more than we can produce electricity from a complex yet powerless circuit. We cannot gain insight of into meaning of a symbol from other meaningless symbols, any more than a function can call itself with the same arguments to discover what it will return and return it.
I deny that meaningless numbers and symbols can produce meaning. Numbers and symbols do not and cannot have meaning without context, and that context must not consider of other meaningless numbers and symbols.
Let’s look at the first claim, that “meaningless numbers and symbols can produce meaning”. We have visual processing abilities such that we can form an image of things that we see. Suppose I see a rock. There is no a priori reason why such an object should be called a “rock”, “fels”, “petra”, or “kklagrh”. I can associate an arbitrary symbol with the visual image. Furthermore, I can communicate this arbitrary symbol to you. I can point to the object and say “kklargh”. Now we have a symbol and shared meaning. Lather, rinse, repeat.
Furthermore, this is exactly how mathematics is defined. Euclidean and Non-Euclidean Geometries states, “Also, we cannot define every term that we use. In order to define one term we must use other terms, to define these terms we must use still other terms, and so on. If we were not allowed to leave some terms undefined, we would get involved in infinite regress. Euclid did attempt to define all geometric terms. He defined a “straight line” to be “that which lies evenly with the points on itself.” This definition is not very useful; to understand it you must already have the image of a line. So it is better to take “line” as an undefined term. Similarly Euclid defined a “point” as that which has no part–again, not very informative. So we will also accept “point” as an undefined term. Here are the five undefined geometric terms that are the basis for defining all other geometric terms in plain Euclidean geometry: point, line, lie on, in between, congruent.” The terms are undefined, yet we form a visual image of them, which we can communicate via other terms, for which we also have a visual image. In this way, we create and share meaning.
In Introduction to Artificial Intelligence, Jackson writes, “A mathematical description of something consists of a finite set of statements (axioms) that utilize a finite set of undefined terms, together with a finite set of rules that govern the derivation of new statements from the axioms and from previously defined statements. Such a collection of statements is called a mathematical system, or theory, and the concept is that any statement, either given or derivable, is a true statement concerning the thing described by the theory. A mathematical theory may thus enable one to use a finite number of statements to describe something about which an infinite number of statements (those derivable under the theory) are true.
For example, the mathematical theory of Euclidean geometry gives us certain axioms or postulates concerning the undefined concepts of “point,” “line,” “plane,” “between,” etc.; the “thing” described by this theory is a “geometry,” consisting of interrelationships existing among lines, points, planes, circles, spaces, etc.
The ingredients of a mathematical theory, then, are the following:
1. A set of basic words (e.g. “point,” “line,” “between,” “distance,” “x,” “y,” “not,” “implies,” “for all,”) that refer to different objects, relations between objects, variables, logical connectives, quantifiers, and so on. These are the undefined words or symbols of the theory.
2. A set of basic sentences made of these basic words. These basic sentences are the axioms or postulates of the theory.
3. A set of logical rules, also made of these basic words, that tell us how to derive new statements from the ones we are given.”
Given this, the second claim, Numbers and symbols do not and cannot have meaning without context isn’t strictly true. “One” means “the successor to zero”, regardless of whether it’s one apple or a bit in a 32-bit integer.
The third claim, context must not consider of other meaningless numbers and symbols is obviously false, as shown by the previously cited works concerning how undefined terms are joined together to produce meaningful systems.
I can associate an arbitrary symbol with the visual image. Furthermore, I can communicate this arbitrary symbol to you. I can point to the object and say “kklargh”. Now we have a symbol and shared meaning. Lather, rinse, repeat.
Indeed. But try doing it without using the term “I” or any of its equivalents to conjure up the symbols.
And consider too the distinction between symbols and signs.
Indeed.
So does this mean I successfully answered Gigalith’s objections?
But try doing it without using the term “I” or any of its equivalents to conjure up the symbols.
You moved the goalpost. Nobody said hard AI was easy. But I think I’ve countered the representation problem (which I don’t think I could have done were it not for orthodox Christianity), and the production of meaning objection. I can also handle the morality problem, but I’m going to save publication of that on my website. I don’t have any answers for the self-awareness problem. Hofstadter tried to tackle it in Godel, Escher, Bach; but I don’t have any special insight into whether or not he’s on the right track.
And consider too the distinction between symbols and signs.
My brain is fried. And Google gives a number of answers, i.e. “Signs have meanings and symbols make you feel something.” and ” According to Tillich and Jung, a symbol, in some manner, partakes or participates in the reality towards which it points, whereas a sign simply points to something greater.” What distinction(s) do you want me to consider?