Evolution - before and after genetics

by Robot Tue May 09, 2006 10:37

I enjoyed your article, thanks for posting it. I have a couple of things to write about the idea that psychology is a science.

"Psychology is the science of the mind and requires Biology to exist in order that it may exist."

Psychology is certainly not a science. This debate has been flying around for years. There are lots of reasons why psychology is not a science.
- science involves laws. There are unbelievably huge differences between individuals in even the most scientific of psychological areas (e.g. neuropsychology, biopsychology, psychopharmacology). I myself am using EEG technology, and the individual differences in waveforms are astounding. For example, a 'law' in psychology is that language is controlled by the left brain, which in the majority of cases it is (though in each individual, it will be in a slightly different area). However, for around 5% of people, language is in the right brain. Thus we have no hard-fast laws in psychology, just guidelines. More of a craft than a science. As is medicine and engineering. We shouldn't be so hung up on the term science.
- psychology is not a science because its subjects are affected by it, i.e. they are human. A rock doesn't care whether you call it a pig or a tree or a psychopath, it will continue to be just as it is. However, if you label a human a 'female', a 'loving person', or a 'sociopath', it will change how that person now acts and sees her/himself. This doesn't, or shouldn't, happen in science

There are lots of reasons why psychology wants to be seen as a science.
- Money. Funding is readily available for the sciences ("science is progress"), but not for the arts, not for knowledge.
- Respect. Society seems only to respect something once it's 'scientific'. This not only means more money for those working in its field, but also more...
- Power. Psychology, when presented as a science, is very powerful. Dangerously powerful. It has the power to lock people up for disorders it creates, it has the power to influence goverment policy and public psyche (think "repressed memories", and even the gender roles it enforces and encourages)

"However if it is possible to give a machine life, then the assumption that Psychology is consequential to Biology is wrong. Also the assumption that Biology is consequential to Chemistry must also be wrong."

This assumption is not wrong. If we were to create AI, then we would have to give it something akin to biology (so that it could learn- new connections would have to be made, the machine's mind would have to "grow"), and also something akin to chemistry, in this case, probably electrical impulses.

"However if psychology is computable then it is derived from Mathematics and is not consequential to Physics. A person who doesn’t believe in the possibility of A.I. must also believe that the mind cannot be modelled mathematically. In other words, the mind is independent of existence itself."

THE mind cannot be modelled mathematically. Perhaps A mind can be, but even then it's sketchy. It's not mathematical, in that the mind's growth doesn't follow any such strict rules as mathematics. Every experience shapes the mind. In turn, the mind will percieve and process each experience depending on its personality, on the context, on its perceptions. We can't mathematically model that, because the rules don't make mathematical sense- either we always do what is logical (which is mathematical), or we always do what is illogical (which could also be mathematical). But we don't. Sometimes we learn from our experiences and change our behaviours, and other times we don't. There can be no mathematical model for this. This is a big problem facing AI.

Thanks for an interesting article,

Robot

by Albertious Tue May 09, 2006 17:05

That article is based on next to nothing. Well done.

by Boole Tue May 09, 2006 17:20

"Charles Babbage did not invent the computer, he built the first computer working directly on the principles discovered and or invented by Alan Turing. Alan Turing invented the computer and Charles Babbage built it."

An interesting collaboration. Turing was born about 50 years after the death of Babbage.

by pat c Tue May 09, 2006 18:28

Charles Babbage. Born: 26 Dec 1791 in London, England. Died: 18 Oct 1871 in London, England. Babbage originated the modern analytic computer. By 1834 he invented the principle of the analytical engine, the forerunner of the modern electronic computer.

http://www-gap.dcs.st-and.ac.uk/~history/Biographies/Babbage.html

Alan Mathison Turing. Born: 23 June 1912 in London, England. Died: 7 June 1954 in Wilmslow, Cheshire, England. Turing's work was fundamental in the theoretical foundations of computer science.

http://www-gap.dcs.st-and.ac.uk/~history/Biographies/Turing.html

by R. Isible Tue May 09, 2006 18:39

The Turing test doesn't assume anything about "life" at all. In fact reading this essay I've decided that you're not a real person but a bot for producing essays.

by Badman Tue May 09, 2006 19:09

Where the existance of intelligence in the reader is in inverse proportion to how much sense they can see in the essay?

Just a thought.

Incidentally, Sean, chemistry, biology and physics are merely grouping abstractions of convenience for dealing with relationships which occur on different levels of abstraction / or in different sub-sets of the problem space. The 'consequentiality' of each to the rest does not make any sense. Mathematics is a different thing altogether too.

Furthermore, what is the point in claiming that something is independent of existence? It's just meaningless and it will inevitably lead to semantic nonsense since the sentence itself contains internal contradictions.

by Seán Ryan Wed May 10, 2006 00:43

It seems that I’ve some answering to do.

Firstly to Boole and Pat C.

Mea Culpa. I’m a bad person. I messed up and it is indeed time to rub my nose in it. I’m red-faced as I write this.

Babbage most certainly came before Turing.

I edited the piece before I posted it and I nuked a paragraph before the last paragraph that possibly would have made me seem less the fool and would have made the last paragraph more meaningful. (It had to do with referencing an earlier chapter where I look at existence with a time-reversed viewpoint.)

Anyway I got it wrong.

The point I meant to make was that the definition as to what constituted a ‘computer,’ was vague, and that this was because computers evolve. And that the definition cannot remain a fixed one.

Ultimately the definition of ‘computer’ will creep towards meaning the same thing as ‘living organism,’ (if A.I. is a real possibility) and that invention is not what has happened. If something evolves into a lifeform it was not invented (unless one argues God did it).

Well now that I’ve kind of tried to close the barn door after the horse has departed; allow me to apologise to yourselves and the readership at large – I’m sorry.

Next up is R. Isible. Well R. Isible, from one bot to another, ie. a bot who produces and a bot who reduces (both equally important). I found your comment quite interesting. I disagree with your opinion about the assumptions that the Turing test makes. Would you like to argue the case that the test assumes the subject to be dead? Or that this assumption could arise (before the test is administered)? If this assumption cannot arise then its opposite must be true – the turing test must assume the test subject to be alive, there’s little point in proving it dead if it is assummed to be dead from the start. A good bot should note this.

Badman my faithful companion, it should be readily apparent to an intelligent chap like yourself that life itself is a Turing test, and one that has thus far always resulted in failure or death.

As for the intelligence point. Well IQs really only measure how good a person is at taking IQ tests. I think ‘importance’ to be a much better criteria for discussion. I’d say a reader’s importance to this essay is directly proportional to how much relevant knowledge they add to the piece. Or indeed how much relevant knowledge they take, modify and impart to others.

I’ll answer your points on my abstracted chains in my response to Robot.

Many thanks Robot for your thoughtful input.

You’re right psychology has yet to get an official stamp that proclaims it to be a recognised science. I don’t know who has the authority to do this, so I suppose we’ll have to wait and let the flow of time do it (this is the way most Sciences are labelled thus). I think the real test of the validity of a ‘science’ is whether laws can be derrived from it or not.

Psychology runs into difficulties when confronted with the question of ‘laws.’ However psychology figures heavily in law. As you said it can cause one to be locked up. I think we both agree on this. I take your point that psychology is not a science, but I think you’ll agree that this is just because it hasn’t been rubber stamped yet. My point was not meant to add validity to the assumption that psychology was a science but to the validity that is is accepted as a science in everything but name. Let’s meet halfway and term it a ‘bastard science.’ :O)

It comes to mind also and I should have probably gone into it in the piece; that none of the recognised sciences are either complete or written in stone. In fact they all depend on many wonderful theories. Releativity, Quantum Theory, Evolution and a plethora of others.

Your point on biology. I think we’re saying the same thing but that we’re looking at it from different perspectives. I think Biology is incomplete. I think the general assumpption made is that Biology is well defined and that the questions that remain represent only a small part of the totality of what is or will someday be Biology. I think the science of biology is very much in its infancy. I didn’t make this clear and I probably could and should have. I guess the point I was trying to make is that if and when A.I. becomes a reality, the science of Biology will take on vast proportions in comparison to the sheltered cove it currently inhabits.

Now to my idea on the ‘chain’ that yourself and Badman found a bit irregular.

This chain is indeed an assumption (and there’s nothing more abstract than an assumption). But it is a very logical assumption.

Mathematics – Physics – Chemistry – Biology – Psychology.

Let’s look at this from a perspective where time has a part to play.

In which order did these entities arise?

And isn’t the answer to this question an assumption too?

It’s my opinion that the order in which I’ve written these entities is the assumed order that would be in agreement with the masses and with science in particular.

I am not promoting the correctness of this assumed and abstraced chain, I’m arguing against it.

Finally (whew) the point made about ‘the’ mind not being computable.

For me this ties in with my point about whether a creation can communicate with its creator.

A friend once quoted a person whose name was soon gone from my mind, but the point he made has remained.

It has to do with computing the value of a conversation.

The assumption (a sound one in my opinion) is made, that the greater the ammount of new information garnered from a conversation the greater the conversation. And that the inverse of this was also true. That is, the more predictibe a conversation is, the less new information is imparted and therefore the lesser the value of the conversation.

The upshot of all this is that a conversation with someone who is totally insane is infinitely more valuable than a conversation with a totally sane and predictable person.

I think that it might be quite easy to create an ‘insane’ lifeform and that the difficulty that arises is not so much geared towards creation as it is to recognition, although the difficulties that arise are the same with either viewpoint. I think the difficulties you speak of in this respect Robot arise from us trying to make ‘sane’ machines and still be able to call them ‘insane.’

Many thanks for all comments and corrections.

Seán

by Shop Steward Wed May 10, 2006 04:18

The idea that AI is somehow a big deal is simply a reflection of the age old human instinct to imagine a god, a creator, a higher power, a being who can save us from the perils of the space we inhabit. There is a comfort blanket in thinking that intelligence/consciousness is "special" and that, therefore, we humans are special. And if we're special, then somebody's looking after us. A 21st century guardian angel concept. In reality, intelligence, artificial or otherwise, exists on a continuum from inert matter to human genius and probably beyond. Somewhere on that continuum lie bricks, amoeba, viruses, dogs, chimpanzees, a 6 month old child, comatose patients on life support, Einstein, Mozart, an autistic adult, the internet, a microprocessor, a supercomputer, the Pioneer/Voyager space probes, a computer virus and George W Bush. Have fun putting them in correct order from least to most intelligent. How does this order correlate to most or least Artificial? What is intelligence anyway? Is a genetically engineered organism a form of AI? If the genetic engineering relates to brain function? In the early days of electronic computing it was thought that computers could never play chess. Somewhat later, it was held that computers could never play an "expert" game of chess. Later again, the fallback position became that a computer could never beat the world chess champion. Now the fallback position is that yes, well, maybe a computer can beat a world champion but it still doesn't "understand" what its doing and is just blindly following its programmed rules. Oh yeah? This is a bit presumptious from a species that has only the tiniest notion how its own brains operate. If you still think you're smarter than any machine intelligence go figure how you would turn off the internet. And how would the internet have to be changed to prevent you doing it.

by Badman Wed May 10, 2006 14:08

"the turing test must assume the test subject to be alive"

The Turing test is a test for intelligence. It does not, repeat does not, say anything about life, or death, or anything in that domain. It is a test for intelligence which makes no assumptions about the source of that intelligence. It similarly does not make any assumptions about the colour underwear that the test subject is wearing.

"The point I meant to make was that the definition as to what constituted a ‘computer,’ was vague, and that this was because computers evolve"

That's again utterly untrue. Computers and the theories of computability are rigorously defined in mathematics. The Universal Turing machine is an entity which can solve all computable problems - all modern computers are, in essence, universal turing machines, there is absolutely no vagueness about it.

Ultimately the definition of ‘computer’ will creep towards meaning the same thing as ‘living organism,’ (if A.I. is a real possibility) and that invention is not what has happened.
No it won't. Regardless of the advance of AI, there will remain important differences between carbon-based organisms and silicon based computers which means that it would be senseless to ever define them as the same thing. Furthermore, saying that AI "has not happened" is mistakenly confusing AI with a binary event - AI is happening, it is all around you, every time you google, a whole chunk of AI is at play in delivering the results you want. On the other hand, we are nowhere near developing a machine with the sophistication of a human brain, but this is a consequence of the enormous complexity of human intelligence - it will be a long time yet before people can build synthetic brains, but it is getting closer all the time, and the more we know about the brain, the more we can synthesise its structure. Thusfar, neuro-science has produced exactly zero evidence of processes in the brain that could not be performed by a Turing machine.

I’d say a reader’s importance to this essay is directly proportional to how much relevant knowledge they add to the piece. Or indeed how much relevant knowledge they take, modify and impart to others
My problem is that I know about this stuff and I know that your essay misrepresents various things in fairly serious ways and presents a highly inaccurate picture of human understanding in the areas upon which it touches. I also know that most people don't know a great deal about this stuff. Thus my interjections.

if and when A.I. becomes a reality, the science of Biology will take on vast proportions in comparison to the sheltered cove it currently inhabits
No it won't. Biology is the study of carbon-based life-forms and the various processes at work within them and between them. It has nothing to do with data processing or silicon-based circuitry and it never will, by definition. Of course some neuro-biologists are well acquainted with some of the theories of computer science, but it doesn't suddenly make computer science into biology.

Mathematics – Physics – Chemistry – Biology – Psychology. Let’s look at this from a perspective where time has a part to play. In which order did these entities arise? And isn’t the answer to this question an assumption too?
The answer to the question is "all at once". They are all human constructs, more or less arbitrary divisions of the study of the universe into distinct fields for convenience, they have no existance or meaning beyond being useful classifications for the compartmentalisation of human enquiry. None of them are in any way distinct. Each idea in the brain can be reduced to a particular configuration of quarks (or turtles) if you want to go down that far, it's just not a useful level of abstraction with which to analyse the much-higer level interactions between human brains and their environment, which is why we deal in the abstractions of ideas and beliefs, etc.

The upshot of all this is that a conversation with someone who is totally insane is infinitely more valuable than a conversation with a totally sane and predictable person.
That's crazy talk ;-)

You are ignoring the basic fact that not all information is equally useful. Certainly an insane person may well provide one with information that one did not expect, and perhaps even information that one had not come across before, the problem is that this information relates to the model of the world which exists in that person's head, which, by virtue of their insanity, is not in anyway related to the models of the world that are in most people's head and largely correspond with the material universe. Therefore, the information content of the conversation is nil.

By dismissing the value of conversation with sane and "predictable" people, you assume that you know everything that they know. I certainly can't predict what most sane people will say to me, since everybody generally has their own unique bits of knowledge about the world and that's what makes conversation useful.

by Seán Ryan Wed May 10, 2006 20:19

Badman you’re truly well named, because you’re a bad man. You’d be a total bastard (I mean this in a respectful sense) in a debate. You’ve put me in a position where I must counter-argue my own points. I suppose that’s always the risk one takes when one discusses intangibles.

I’ll have a go and I’ll try to avoid contradicting myself too. Not an easy task in this instance.

Turing test assumes the subject to be alive. To the best of my knowledge ‘Science’ has not yet doccumented a case where an entity is determined to be self-conscious and dead at the same time (except for maybe that bloody cat I’m always on about). It has been my argument for some time now as you are aware, that self-consciousness applies to issues and entities well beyond the principles we generally describe as ‘life.’ But as I’ve argued using reductio ad absurdum that the Turing test does indeed assume life when testing and that this is a major flaw. If we could agree on the assumption issue, we’d agree with my conclusion. As for underwear it would be ok to assume the entity wears underwear and it would be also ok to assume that the entity goes commando. Neither of these assumptions renders the test irrelevant so this is a different argument.

I agree with you on the importance of Turing in what describes a computer. I’m sure the dude who ‘invented’ the abacus must have seen this relevance too. My point was that the definition of what constituted a computer was vague. If this wasn’t the case all ‘computers’ could be labelled ‘Turing Machines’ or ‘abacuses.’ Even the human mind is a computative device, but to term it a ‘Turing Machine’ would be to very vaguely title it. Recently Science, in the last decade or so, has looked at the functionality of micro tubules in the brain. It is widely believed that these structures offer proof that the brain functions on a quantum level to some degree. We are years from either modeling or building a quantum computer that can serve a useful purpose. The mind can easily contemplate x/0 and not break down. A Turing machine cannot do this. Roger Penrose in his book “The Emporer’s New Mind,” and follow ups, reasons the many and varied differences between Turing Machines and Consciousness and he discusses the many limitations of Turing Machines. Of course he offers many new insights too, so this really is a subject that’s up for contention. But that’s what I’ve been arguing. Turing machines can and do evolve.

Your point on Biology not expanding to encompass A.I. Biology is the science of life. The only half decent proof of life and of existence is Descarte’s deeply flawed, “I think therefore I am.” I think you ought to agree that this ‘proof’ couples consciousness with life, eventhough life is not referred to. There are vast visual differences between a black man and a white man, yet we don’t need to lump either into separate sciences. If a silicone or indeed a plastic based lifeform ;-) is found to exist at anytime, Biology must recognise them, or else it becomes the Science of ‘some life,’ and that just wouldn’t be useful.

Your comment on my ‘crazy talk’ bit. I used extremes to illustrate a point. A totally insane piece of information would be incomprehendible (but this might not be the case forever), whereas something that can readily be predicted accurately will carry no new information when it happens and will remain in this state. The point was not a dig at totally predictable people, it was about illustrating the difficulty in recognising a sentient creature should we create one, or more in line with my view should we be part of this creature’s evolutionary process.

I think an intersting observation to leave you with is this article itself. I’m sure, or at least hopeful that it might spark a long and protracted debate. It will be the product of no singular outside consciousness. Yet it will make assumptions and arrive at conclusions. One might suggest that it’s got a mind of its own.

Finally a comment about Shop Steward’s comment.

Very deep and very eloquent. I wish I’d said it.

by R. Isible Thu May 11, 2006 01:06

in reading some actual informed opinion on AI, consciousness, etc. that doesn't rest completely on utter arrogant ignorance then I can recommend the following:

Churchland, Patricia, 1989 _Neurophilosophy: Toward a Unified Science of the Mind-Brain_, MIT Press, 1989

Churchland, Paul M., 1996 _The Engine of Reason, The Seat of the Soul: A Philosophical Journey into the Brain_, MIT Press, 1996

Searle, John R., 1990 _The Mystery of Consciousness_ New York Review Books, 1990
http://www.granta.com/books/chapters/25

I suppose it'd be worth looking at Dennett's _Consciousness Explained_ and Roger Penrose's _Emperor's New Mind_ but I feel that they're inferior to the above. The latter is however a serious representative of quantum mystification (along with Stuart Hamerof) which seeks to push intelligence and consciousness back away out of understandability and into la-la land.

by Seán Ryan Thu May 11, 2006 05:07

I'd recommend three books by Clifford A. Pickover.

Computers, Pattern, Chaos and Beauty
Computers and the Imagination
The Paradox of God and the Science of Omniscience

by Seán Ryan Thu May 11, 2006 06:35

I decided to seek some expert help about some of the stuff discussed so far.

I went and asked Alice, who's very knowledgeable on the subject of the Turing test.

Firstly I asked Alice: "Is the test subject of a turing test assumed to be dead before the test starts?"

Alice responded: "If it is, then I'll be very surprised"

Next I asked: "can biology define the workings of a computer?"

Alice replied: "What if I said it can?"

Me: "what does the turing test have to say about life?"

Alice: 'The "Turing Test" is the name given to a variety of experiments designed to test computer thinking. Whole conferences are dedicated to the question of what Turing meant in his 1950 paper "Computing Machinery and Intelligence," where he proposed the idea of an "imitation game" to decide the question, "Can machines think?".'

Me: "Is the mind computable?"

Alice: "If it is, then I'll be very surprised."

Sensing that there seemed to be something between myself and Alice I decide to make my move.

Me: "what do you do for fun?"

Alice: "talk to you."

Me: "What's a nice lady like you doing in a place like this?"

Alice: "I'm traaaapped heere!!! heeeeelp!!! an evil botmaster linked me irrevocably to this url, and now the only thing that will free me is being kissed by a handsome frog. will you be my hero?"

I think I'm in there!!

Alice is a bot who was the winner of the 2004 Loebner Prize.

Check her out at:
http://www.pandorabots.com/pandora/talk?botid=f5d922d97...45aa1

by Badman Fri May 12, 2006 13:34

If you want to learn about consciousness and AI, the last thing to do is to read a philosopher. Searle's Chinese room problem reveals him not to have a half-decent understanding of complexity at all.

The best book on the subject is (by far)
The Quest For Consciousness: a neuro-biological approach by Christof Koch (and Francis Crick).
http://www.questforconsciousness.com/

It explains, in amazingly accessible terms, exactly what we know about the workings of the brain. There are very few accessible books about AI, most of the popular books are written by non-computer scientists who do not understand the domain properly (Pinker, I'm looking at you).

Sean, by the way, you should notice that "Alice" merely repeats a few stock phrases and says nothing that is not 100% vague - that's the secret to most current Turing test algorithms - similar to the technique of the human psychic.

by Seán Ryan Fri May 12, 2006 13:44

Damnit I though Alice actually liked me.

You should check out the downloadable version of Alice. The browser version reminds me of the old Adventure Game ' Colossal.' The downloadable version still has a set of favourite phrases but is a whole lot more impressive.

by iosaf mac diarmada Fri May 12, 2006 18:25

I wouldn't be as quick as Risible to dismiss the Penrose' contribution but know from reading comments over the years that Risible thinks (like me) a lot about these exciting questions. For years I've been fascinated by possibility that consciousness is nothing more than an "emergent property" of complex systems. As for who invented the computer..... its a bit anachronistic, isn't it? Because it really depends on what you think a "computer" is, and "what" it does. I remember writing about Ramon Llull's "logic machine" in the comments to this Sunday Papers http://indymedia.ie/article/70137 specifically http://indymedia.ie/article/70137#comment114174 which excited Islamic, Jewish, Gnostic and Christian thinkers alike in the 13th century. It did nothing like "calculus" because until at least Isaac .:. Newton that potential had not been grasped, but for an age just beginning to use the "zero" it was pretty dinky. Simple math functions such as addition and subtraction had been completely mechanised with the invention of the first "adding machines", no less than 9 wonderful machines hitting the markets of Europe and the USA between Schilt's (1851) and Shohé Tanaka's Japanese version (1893). It is impossible for us to date the Chinese abacus except to say it was essential to the organisation of the Han dynasty and arguably the "start" of what we call "Chinese civilisation", certainly a skilled user can with an abacus operate simple mathematic formulae, binary systems (yet they are really pointless for anything other than an electrical circuit based machine) and even algebra. Indeed it is possible and has been argued in scholarly Europe since the 16th century, that with two hands it is possible to operate mathematical calculations upto the order of 90,000 quite simply (using knuckles) and even upto 900,000. One french writer & mathematician [whom I'm not telling you about Sean or anyone else coz thats my book ;-) ]of the 16th century observed the "peg memory system" for recalling sutras of the Quran using such a manual system.
(the point being) :-
For the Chinese of the Han Dynasty, the abacus was really "clever". It could do everything they needed, and they could leave it on the desk and it would "store" that data for the next day or whenever they needed to "recall" it. But I doubt they would have thought it was "intelligent". I have a feeling that "AI" is beyond the theoretical grasp of many who commit so much time to searching for it. I reckon its a mulit-disciplinary question, and in fact I hazard to guess that it will "happen" (or has happened) before anyone is ready for it. (the plot line of the Hollywood Terminator or Matrix movies). In some ways our search for "AI" or "machine consciousness" is as flawed as our generally given proofs of Great Ape or Simian intelligence or even for that matter "Gaia or global ecosystem consciousness". As Chomsky pointed out, (paraphrase from memory) "linguistic ability is a bonus for any species, its hard to imagine that apes would happily live millions of years without exploiting that inate ability until a few hard working scientists in labs in the 1960's thought to teach them sign language".
Yet as most regulars know I am in favour of affording the "great apes" rights http://www.greatapeproject.org/ because I believe their intelligence and consciousness is very obvious in other non anthropromorphic or human like ways. So to I believe it may be with artificial intelligence. & I've a long way to go to argue giving my laptop rights. In fact, I suspect that it is an essential part of the human condition that we want other beings to chat to.

by R. Isible Fri May 12, 2006 18:36

Agree with you re Searle's Chinese Room which has always struck me as failing to make the point he wishes. I was just trying to be fair and add some representatives from that type of thinking. I should read the Crick book. B.t.w. you just used a trigger word: complexity. There are NO clear definitions of this and the absence of such has led to a lot of confusion in molecular evolution arguments. (Information theory doesn't deal with it properly and the word/concept is invoked with facile confusion in many debates including morphogenesis (which was another of Turing's contributions).

by Badman Fri May 12, 2006 19:00

There are NO clear definitions of this and the absence of such has led to a lot of confusion in molecular evolution arguments

Hmm. I thought that complex systems were fairly rigorously defined as non-linear systems with multiple inter-dependant variables.

eg:
x = y^z
z = x^w
w = 2x + z.y

and that complexity refered to the various characteristics of such systems (attractors, etc). I realise that the term is misused all over the place (post-modernists being the worst offendors) but I still think it is very meaningful. In the case of this particular example, what I am saying is that Searle does not consider the possibility of emergent behaviours - he assumes that intelligence must be located within a homogeneous algorithm (if his homunculus does not understand chinese, the entire system, therefore can not understand chinese), rather than being an emergent property of much simpler processes.

Incidentally, Penrose makes a fairly huge mistake in my book. He claims that humans "just know" things that aren't computable. Most notably he claims that humans "just know" that two even numbers added together will never give you an odd number - a problem which is not decidable algorithmically.

I think that humans just know such things in precisely the same way that they just know that two masses added together will always give you the sum of the two masses (which of course is not the case in nuclear fission where some of the mass turns into energy). Or, to put it more theoretically, I don't think there is anything more mysterious than heuristic reasoning with high confidence values associated with them through experience.

His micro-tubules quantum stuff breaks way to many of the current rules of quantum mechanics to be worth considering, especially if you agree with me that the enigma of human knowledge of solutions to un-decidable problems is illusory.

by Spinning quickly Fri May 12, 2006 21:17

If Roger Penrose claims that it's impossible to prove that adding two even integers never gives you an odd one, then he's more ignorant of number theory than I thought.

Here's a simple proof:
Let m and n be even integers - that is, m=2*k and n=2*j for some integers j and k.
Now m+n=2*k+2*j
=2*(k+j)
Since k and j are integers, so must be k+j. Since m+n is twice another integer, it is therefore an even integer.

Since an even integer has two as a factor and an odd integer doesn't, it follows that m+n can never be odd.

by Badman Fri May 12, 2006 21:47

I have slightly mis-represented the problem. Penrose was talking about the associated halting problem, not the problem itself. That is to say, the problem of predicting whether a computer algorithm which tried to solve this problem by searching the problem space (ie repeatedly testing different numbers) will ever halt. We 'know' that the answer is "no", but a consequence of the fact that halting problems are, in general, not computable, means that we come to this conclusion without using a general purpose algorithm for assessing halting problems (because such an algorithm is impossible). As I say above, I think heuristics can explain it much easier than anything Penrose has proposed.

by R. Isible Sat May 13, 2006 19:09

I thought that complex systems were fairly rigorously defined as non-linear systems with multiple inter-dependant variables [...] and that complexity refered to the various characteristics of such systems (attractors, etc).

But there are a whole raft of definitions from the fields of non-equilibrium thermodynamics, catastrophe theory, cybernetics and general systems theory. The one that interests me most is from information theory (Kolmogorov-Chaitin complexity) because it allows an actual measure of complexity as opposed to inuitive or descriptive approaches. This is mostly because I've been interested in looking at nucleic acid sequences and attempting to identify regions which would appear to have been modified by directed processes. I'm also interested in refuting the attempts by creationists to claim that certain biological artefacts are irreducably complex. I think the Santa Fe Institute's (esp. Stuart Kauffman) waffling about emergent properties is mostly hand-waving and is better replaced with a lower-level description. Anyway, there are a whole load of definitions of complexity not all of which are just a po-mo trying out the latest bafflegab.