Barry has a post up at UD that is on the same topic as one that I’ve had half written for a while now, but I thought I’d jump the gun and comment on Barry’s here, as it raises an important point, nicely and simply made: that, as Barry’s post-title puts it:
“If my eyes are a window, is there anyone looking out?”
Barry writes:
As we were winding our way through Custer State Park I became aware of myself looking through my eyes as if they were a window. I had a keenly felt sensation of what theorists of mind call the “subject-object” phenomenon. I perceived myself as a “subject” contemplating and having a reaction to an “object” (the beautiful scenery of the park).
Given their premises, materialists must believe the brain is a sort of organic computer, in principle very much like the computer on which I am writing this post. The subject-object problem is a seemingly insurmountable obstacle to this theory. Closely related to this issue is the idea of “qualia,” the subjective perception of experience (the cool blueness of the sky, the sadness of depression, the warmth of a fine sunset, the tangy-ness of a dill pickle).
Consider a computer to which someone has attached a camera and a spectrometer (an instrument that measures the properties of light). They point the camera at the western horizon and write a program that instructs the computer as follows: “when light conditions are X print out this statement: ‘Oh, what a beautiful sunset.’” Suppose I say “Oh, what a beautiful sunset” at the precise moment the computer is printing out the same statement according to the program. Have the computer and I had the same experience of the sunset? Obviously not. The computer has had no “experience” of the sunset at all. It has no concept of beauty. It cannot experience qualia. It is precisely this subjective experience of the sunset that cannot be accounted for on materialist principles. It follows that if materialist premises exclude an obviously true conclusion – i.e., that there is someone “in there” looking out of the window of my eyes – then materialist premises must be false.
The question in the title of this post is: “If my eyes are a window, is there anyone looking out?” The materialist must answer this question “no.” That the materialist must give an obviously false answer to this question is a devastating rebuke to materialism.
So, first of all:
Given their premises, materialists must believe the brain is a sort of organic computer, in principle very much like the computer on which I am writing this post. The subject-object problem is a seemingly insurmountable obstacle to this theory.
It is true, in a sense, that I think (“believe” is not a word I find very useful – “posit” would be better) that “the brain is a sort of organic computer”. It is certainly organic, and it certainly computes things (in my case, not very well, which is why I use a computer!) But I do not posit that my brain is “in principle very much like the computer on which I am writing this post”. If it were, then the “subject-object problem” would indeed be “a seemingly insurmountable obstacle to this theory”.
For a start, the computer on which I am writing this post receives all its input from human sources. My brain, in contrast, receives its inputs from all manner of external sources, and, what is more, depending on those inputs, “computes” a motor response which it sends to my body (my eyes, my neck, my torso, my legs) that changes the input. In other words, my brain is not (or not simply) a tool of some other intelligent agent, my brain is the “tool” of the organism that I call “me”, and which incorporates (literally incorporates) not only my brain, but my entire body, motor and sensory apparatus, digestive, circulatory and endocrine system and all. So the brain is not simply an information-processor, like the computer on my desk, but part of an information-gathering system – moreover, one in which the information to-be-gathered is itself an output of the system.
Secondly, as implied above, this makes brains a subsystem of a whole system that is most strongly characteristics by re-entrant feedback loops, in which not only is information processed, but in which the output of that processing is re-entered as input into the decision-making process as to what further information to seek. So if we want a materialist analog to the brain, we need to look at robots, not computers – i.e. things that can move their sensory apparatus as a function what information they need.
Ah, need. That’s another thing – organisms have needs (at its most simplest, to survive, but with all kinds of sub-needs, and epiphenomenal needs supporting that basic need – we should probably leave the origin of those needs to one side for now…) Organisms have needs, therefore they potentially have goals – outcomes that they seek, which we can also express as “desire, and take action to fulfill”. And those goals themselves are part of what the brain sets, and changes, in the light of new information.
So no, Barry. I, as a materialist (and, it should be said, a cognitive neuroscientist!) do not “believe” that the brain is merely an “organic computer” that is “similar in principle”, to the one on your desk. I think it is radically different to the computer on your desk, not least because it does a heck of a lot more than “compute”. It is part of the system of tools that enables me, as organism, to survive, by not only computing what I have to do to reach those goals, but to compute those goals themselves, in the light of current information, to seek further information in that may result in further adjustment of those goals, and to select actions that will enable me to fulfill them.
So….
Closely related to this issue is the idea of “qualia,” the subjective perception of experience (the cool blueness of the sky, the sadness of depression, the warmth of a fine sunset, the tangy-ness of a dill pickle).Consider a computer to which someone has attached a camera and a spectrometer (an instrument that measures the properties of light). They point the camera at the western horizon and write a program that instructs the computer as follows: “when light conditions are X print out this statement: ‘Oh, what a beautiful sunset.’” Suppose I say “Oh, what a beautiful sunset” at the precise moment the computer is printing out the same statement according to the program. Have the computer and I had the same experience of the sunset? Obviously not. The computer has had no “experience” of the sunset at all. It has no concept of beauty. It cannot experience qualia.
Indeed, your computer cannot. That is because your computer is not an autonomous interacter with its environment, in which it controls and adapts its own goals according to its needs – indeed, it has no needs. We may need computers; the computer does not need itself – it does not need to survive.
It is precisely this subjective experience of the sunset that cannot be accounted for on materialist principles.
And so we have the non-sequitur:
- P1 Materialists think brains are computers
- P2 Computers cannot have experience
- C: Materialist principles cannot account for experience.
Not only is the first premise wrong (see above), but C doesn’t follow anyway, because experience is not simply a function of brains but of entire organisms.
So this is wrong:
The question in the title of this post is: “If my eyes are a window, is there anyone looking out?” The materialist must answer this question “no.” That the materialist must give an obviously false answer to this question is a devastating rebuke to materialism.
My answer is no, not because I think there is no-one “looking out” but because I don’t accept the premise that “my eyes are a window”. My eyes are not a window, they are simply the things I (qua organism) use for looking with, and I don’t “look out” of my eyes – I just look.
So there’s certainly someone looking.
Who it is will (probably) be the subject of my next post 🙂
Guys, check the rules – some stuff moved to Guano. As ever, no moral judgement is implied 🙂
And William has reminded me I’ve been getting slack lately….
I see KF has added 4 PSes to his post. I aslo see that his wish to have posts he would find objectionable moved has been granted. I’ll simply note, he finds a lot objectionable.
Tonetrolls 1, Billygoats 0.
While I realise we have a somewhat odd symbiosis with UD, I don’t want this site to become a peanut gallery for UD. There already is a site for that 🙂
Ideally, it would be nice to have conversations with UD regulars in person, but things being what they are, loudhailing from site to site seems to work quite well.
But I’m not going to moderate this site according to anyone else’s rules but my own.
Oddly, he hasn’t linked to this thread, but to a post of Mike’s in sandbox.
It would be nice if he actually read this thread.
We’ve been asking for direct dialogue for ages. UD is a waste of time, comments don’t get through, get, changed, get deleted, get promoted after the conversation has finished. KF issues challenges in a venue he controls. He also wished you to stop people saying things *he* finds objectionable.
Personally I’m against censorship, and pro dialogue. I also believe its okay to hold opinions about people and things if you are genuine and can explain why you hold them, and are prepared to subject them to scrutiny. I have explained why I find IDists to be the ‘American Taliban’, a moniker they doubtless don’t enjoy but I feel is warranted, and will defend.
ID has explicitly stated its aims and objectives. And we don’t forget.
Byers, so we could remove the brain and the subject would still be able to think?
What about people with massive brain tumours? Is their thinking unaffected by what is wrong with their bodies? And Alzheimer’s disease – how do you explain that, Robert? Being tired doesn’t affect our thinking? Being drunk, we can still think perfectly well?
Look up Phineas Gage and tell me your thoughts on the subject. I am fascinated to know what you think.
How does being drunk work under Cartesian dualism?
You do not understand what is under discussion. Here’s a simple program having fewer than a hundred lines of code. The code is as deterministic as you could wish, but you cannot predict its output. I bet my house you cannot predict its output.
A program with a sense of beauty
Just click on start.
GIve me the code and enough time I will tell you the output.
It has a stochastic component, so I doubt it. If you’ve cracked that, the stock market beckons.
It’s not possible even in the general case, Blas. See the Halting Problem for more details.
My favorite proof is here.
Of course it has a stochastic component. Given the resources it could use something like radioactive decay for its source of random input.
But the output is not random. It has a definite tendency to produce “beautiful” strings. You just can’t predict what they’ll be.
Not understanding that is a fundamental failure to understand what this thread’s about.
Richardthughes,
You are talking about Deep Blue or Petrushke example?
As far as I know Deep Blue process complete games of each possible move and filters according an algorithm the winners, no stochastic there or at least very few stochastics. The more stochastics in chess game programs the more they lose.
With Petrushka example, yes it is stochastich I cannot predict one output but give me enough time and I will predict all the possibles sequences of outputs. It is a big number but not infinite, at least in spanish are combination of 28 letters. And for the point in discussion, stochastics programs are not going to produce AI.
petrushka,
I immagine that “beauty” is defined by a non spontaneus algorithm.
I would imagine that beauty is a result at least in part of appetitive algorithms.
Correct. Beauty in this algorithm is defined as letter combinations that occur in dictionary words. But the program can and does make words that are not in the dictionary. That is because only a small fraction of “beautiful” combinations are in use.
But being able to list all the possible combinations is not a prediction. I can list all the notes on a piano keyboard without being able to predict what someone will play.
Yes, as explained above. But in this case the appetite is very general. At its most fundamental a preference for letter pairs that occur in dictionary words. Triplets preferred to pairs, and so forth.
I originally intended this as a game between humans. The objective being to evolve words using artificial or, as gpuccio might say, intelligent selection.
I added a Markov selector to make the demo, and the demo seems to me more interesting than the game. It reveals very well hidden connections between what appear to be isolated islands of function.
That would be wrong.
“• A opening book of about 4000 positions was used, as well as a database of 700,000 grandmaster games from which consensus recommendations could be extracted.
• Iterative deepening alpha-beta minimax algorithm with transposition tables, singular extensions and other chess-specific heuristics and a historical database of opening and end games was used
• Generates 30 billion positions per move with average depth of 14 routinely”
Deep Blue – search algorithms.
Note especially the second and third items.
I noticed that, too. I mean, I knew that I’d exploded in anger, but I thought I was being so discrete about it!
Alas, it’s the standard KF version of the Gish Gallop — the same old quotes trotted out again and again, without context or interpretation or analysis (Haldane? Check. Provine? Check. Plato? Check.) And of course KF can’t be bothered to give us any presentation of any metaphysical naturalists who accommodate rationality, agency, or moral responsibility within the natural world. Say, Dennett (“Freedom Evolves”) or the Churchland (“Plato’s Camera,” “Braintrust”), Flanagan (“The Problem of the Soul”, “The Really Hard Problem”), Thompson (“Mind in Life”) or de Waal (“Primates and Philosophers”). But that would involve the excruciatingly painful work of becoming informed about a subject-matter before expressing an opinion about it, and who can be bothered to do that these days?
No, it is not a prediction as your program is not intelligent there is no way a computer program will be.
Give me an example — not involving sex — of an intelligent task that computers will never do. Preferably some job that humans get paid to do.
Are you familiar with the story of John Henry? Let’s have a hypothetical contest.
Blas,
The number of possible chess games has been estimated at 10^(10^50). Even if every subatomic particle in the universe were a supercomputer working in parallel, it would be impossible to “process complete games of each possible move and filters according an algorithm the winners.”
Following your example of the piano, write “impromptu”.
Cibernetic music is awful.
Fair enough, even though you have chosen a task which fewer than one in a million humans can do well. My question specified never, but I suspect the future can be foreseen in the work of David Cope. Try googling and listening to some of his stuff.
I’m willing to bet that not one human in a hundred can reliably distinguish the output of his program from that of the original composer, and he is a pioneer.
Computers and programs generally get better over time.
As for it being “awful.” many people think much of the modern concert repertoire is awful, particularly 20th century “classical” music.
I would like to link to one or two of Cope’s pieces, but I am at a computer having no speakers.
Deep Blue does look about seven moves ahead (14 ply), but the question is how does it determine which ending position is the best. It can rather easily eliminate obvious losers, but that still leaves a vast array of game positions. As anyone who plays knows, one cannot simply count up the pieces and see who’s winning.
Agree with you, good music died with Stravinsky.
Other task a computer will never do is, as I said in a previous comment directed to Lizziem, abstraction. Create a new “concept”, like humans that could define the mathematical circle without have seeing it ever.
There will definitely be sex bots.
Who needs modern classical music when you have Brian Wilson?
Actually, I personally think Gershwin and Ravel wrote great pieces of modern classical music. Charles Ives I like too. I suppose technically they are not post-Stravinsky. The work of Bernard Hermann, John Williams, Michael Nyman and other film composers is none the worse for being commercial, imo.
A quick search for “machine generated music” gives a lot of hits, including one describing a case where members of a concert audience couldn’t tell the difference between human and computer generated music. Searching “computer generated pop music” shows that to be an area of active research as well.
I don’t think you picked a good example.
I didn’t mean to imply I agreed with people who don’t like modern composition. In fact Stravinski and Satie are my favorite composers. Along with Mozart.
Your argument regarding the circle is why I asked you to name a task that educated people get paid to do that will never be done by computers.
The simple fact is that human geniuses are rare, and as computers began to take over intellectual shovel work (as machines took over physical shovel work) they will begin with repetitive tasks and advance toward tasks that require broader knowledge and judgement.
I think you are dead wrong about the arts. I suspect there already machine compositions that have been sold for money. I’m not sure anyone would want to admit to it just yet.
Computers are rapidly taking over the onerous task done by “in-betweeners” in animated movies. Computers are converting movies to 3D. You can even buy television sets that do this on the fly.
None of this is slowing down. I’m asking you to name a task (a job) that cannot ever be done by a computer. Being a genius is not a job. Even if it were I’m not convinced it can never be done by a computer.
There are patentable electronic circuits created by genetic algorithms. Perhaps not works of genius, but very interesting.
“Create a new ‘concept’, like humans that could define the mathematical circle without have seeing it ever.”
This sounds interesting, but I’m not sure exactly what you mean. Are you referring to only mathematical concepts of some sort?
That means that computers are intelligents? Weren´t that computers programmed with algorithms that chose strings of music of defined caracteristics nice to humans?
I will wait until a computer´s composition gets de history of music.
Abstraction is not a genius work, everybody can immagine and define non existan objects. All the words of every alphabet is an abstraction.
Two jobs that will not be performed by bots: stockbroker and general of army.
Patrick,
“Needless to say, the digital Debussys weren’t doing this on their own. As Eigenfeldt and his colleagues note, the compositions reflect “the artistic sentiment of their designers.” The approval of the audience suggests the systems in question “were successful in portraying the goal, aesthetic and style of the two composers who generated them,” they write.”
Most stock trades are done by algorithms. Probably 90 percent.
As for generals, you are talking about whether humans will relinquish control; not whether computers could do better. I think you are way behind the curve on both of these.
Perhaps you simply don’t understand how progress works. Machines did not simply step in one day and take over all manual labor. They evolved. In the last ten years, most assembly of mass produced items is done by robots. The percentage just keeps going up.
Perhaps the job of creative artist will be among the last to go.
Human composers try to get the approval of their audiences, too. What’s the difference?
Can you articulate exactly what it is you think that a human can do that cannot be replicated in hardware and software, regardless of advances in the state of the art?
Yes I know, algorithms made by humans and parameters set by humans. No human will put his money in the hands of a self programmed bot.
Computers are not going to do better than the human that programmed it. The only thing that computers do is process massively information and contrasting it against parameters prefixed, or variable according to prefixed conditions and information processed. Then they give the answer that the programmer were looking at, and would give if would processed all the information.
Who do not understand how progress work is you. Machines didn´t evolve. Humans design better machines that solve better humans problems.
Humans want, and immagine the machines. Machines can´t and never will.
Will and abstraction.
OT: Can anyone help me how keep the answers in the same box. Thanks.
I think we need to keep in mind that ten years ago, just the idea of being able to emulate a known composer would have seemed ludicrous.
I will make a bet that within five years there will be a commercial app that will compose music using immediate EKG feedback from a human “audience.” Will it be a machine composition? How would you even define the question?
We are entering a world foreseen in the movie Forbidden Planet, where anything we imagine can be created by a 3D printer.
What they failed to foresee is that the imagining will be interactive. Our imagination will be amplified and extended by computers.
They already do.
What you will increasingly see are self-programming computers. They already exist. At some point they will be able to improve themselves.
And how do they get the instructins in order to self programming?
I suspect that sporting competitions will never be performed by bots. What would be the point? For instance, Robotic Grand Prix Horse Jumping would kind of take the excitement out of the event.
Click on the gray Reply box. That will keep your response attached to the original.
I think we have some clear differences of expectation established. I don’t think anything more can be accomplished by arguing that my vision of the future is bigger than yours. It’s been fun.
Bu I think it needs to be constrained by the need to list things that people do. Will and abstraction are not operationally definable. The production of music and art are tasks that can be defined, so I accept them as good candidates for things difficult to do by machines.
In ten or twenty years we can revisit these predictions.
You mean, what’s the origin of artificial life? 😉
You’ve never seen BattleBots?
Considering our love of mayhem, and the increasing concern for sports injuries, I think that extreme bot wars are an inevitable genre.
I think we already have the movie franchise.
“Will and abstraction.”
Excellent! Seriously, I appreciate the direct answers.
Of course, now we get to the part of the discussion where I ask for operational definitions of those two terms. How would I know whether or not a software system exhibited “will”? How would I know that the output of a software system was an “abstraction”?
“Can anyone help me how keep the answers in the same box.”
You have to use the small grey “Reply” button underneath the blue “(Quote in reply)” and “(Reply)” links.
I’d watch in anticipation of the energetic failure modes.
At the moment, the automated trading algorithms I’m familiar with use large amounts of historical data to discover patterns. Some use biologically inspired mechanisms, neural networks in particular. Those frequently involve some form of feedback that makes the system self-modifying.
There’s also been a lot of interesting work based on Koza’s genetic programming, which evolves full programs as opposed to patterns and predictions. GP can create new instructions.
That’s why I tried to limit the discussion to things people do for money (jobs). Computers are doing for intellectual work what machines and power tools did for manual labor. They leverage what people can do and extend our power. The same thing is happening with mental labor.
Machines are extending our ability to think, and in doing so they are acquiring abilities that were not anticipated.
I’m pretty certain that at some point they will take over their own design.
My group uses http://creativemachines.cornell.edu/eureqa