Uncomputable Insight

My Turing Test post turned out more interesting than I expected. Playing iGod was fun, and just that; but Steve’s link to a musical Turing Test really surprised me. Looks like speech is more quintessentially human than music … this is a challenge to those who see Mozart as a higher art form than Metal. Bach may be easier to fake than Beatle’s lyrics.

Meanwhile I finally finished “The Emperor’s New Mind”. As various folk said, here and privately, although its a fantastic overview of various parts of science, the “consciousness as quantum gravity” line does not end up being convincing. However .. the basic idea that human minds do something uncomputable is intriguing. Its closely linked to an issue that bugs me about mysticism, and especially Taoism and Zen Buddhism.

There are some aspects of Zen/Tao that fit well with the scientific outlook. Mysticism is not misty and vague, but rigorously insistent on the physical world, as opposed to the mental constructs we mistake for the world. (Including the fairy tales the West calls “religion”.) A classic Zen lesson is to hold up a book and ask the pupil what it is. “A book” says the pupil. “No” says the Master, “book is a noise.”. He plants it firmly in the pupil’s hand. “This is what this is”.

Anyway, what Zen and Science have in common is the belief that there is a real concrete world, and that there is a route to knowledge of the world. In both cases we carefully observe the physical world. But beyond this the (official) methods of Science and Zen diverge. In Science the route to truth is a kind of loop around guessing, testing, and refining, together with eternal scepticism. In Zen, there is first a kind of Brechtian alienation, to shock the mind out of false assumptions; but this goes with a belief that once you do that deconstruction, you already know the answer. You meditate on the flower until “The Flower” fades from your mind and your eyes see … the real flower.

Hopefully you get the link back to Penrose. Zen says that you can know the truth all-at-once, that the mind can do this. Science ..or at least textbook scientific philosophy – says you can’t. But of course many scientists also place great stock on insight, intuition, creativity etc. And push a little harder and scientists split into positivists (its meaningless to ask about essential truth; you can only know what works) and Platonists (we may be wrong at any stage, but our aim is to describe reality).

Can we know the world, or is our knowledge ever provisional ?

Advertisements

13 Responses to Uncomputable Insight

  1. I think most people agree that Penrose’s books
    (do read them all) offer a good overview of physics
    and that the quantum consciousness stuff is wrong.
    More interesting are his ideas on quantum gravity.
    Penrose is in a minority in that he thinks that
    quantum mechanics must be modified in a fusion of
    QM and GR. He put his (or God’s) finger on the
    initial-entropy problem decades before it was picked
    up on by more mainstream physicists. The last I heard
    was that he was working with Zeilinger on actually
    experimentally testing his ideas on QM—no
    ivory-tower theorist he!

    Also, check out the two books by Robert Pirsig:
    Zen and the Art of Motorcycle Maintenance and
    Lila.

  2. Tony says:

    “You meditate on the flower until “The Flower” fades from your mind and your eyes see … the real flower.”

    But that is all you see: the outward physical appearance of the flower. Seeing nothing but that means that you are emptying your mind of all known links to the term ‘flower’: all knowledge of the science and literature of flowers, all images of past flowers, all the emotions that the sight and term evoke. Ridding oneself of all that may lead to a sort of peace but it seems to me an enormous cost to pay.

    I suppose it is possible to speak of this as a purer form of knowledge but it is a form shorn of any possible links to anything else. It cannot lead to the creation of anything else. Unlike science or literature, it is empty of all meaning and so cannot be the source of creation. I prefer my knowledge to be such that it leads to new knowledge, new ideas, new technologies. I find this huge web of knowledge to be truly beautiful and the single empty flower rather desolate.

  3. andyxl says:

    Phillip – back in the seventies “Zen and the Art” was briefly so popular we got free copies with our student ID cards. Or was that “Court and Spark” ? Hmm. Never got round to “Lila” though.

    Tony – perhaps you don’t need to conclude that all those ideas and sense impressions concerning flowers are not real – just that they are not the flower. Then the concrete flower doesn’t seem isolated and depressing. Also … its not supposed to be isolated. At your moment of enlightenment you see it not even as an object, but more of a process. There is no flower, only the Universe flowering. Just one node of the jewel-net of Indra. Jeez I sound like a Zen salesman ! Stop me before I start seeing Buddhist Angels.

    By the by, don’t you think “The Concrete Flowers” sounds like a great name for an Art Rock Group ?

  4. Steve W says:

    Zen, schmen. Rather like those beautiful insights you have just as you are falling asleep. Mine at least are gibberish, as a couple of times I have managed to force myself awake and analyse the grandes pensees – complete twaddle they were.

  5. andyxl says:

    Steve – I thought you made your own sushi ? Surely you need Zen concentration to do that right ?

  6. Tony says:

    Ah, the zen of sticky rice.

  7. John Peacock says:

    Andy: I find the metaphor of the shadow helpful in addressing the question of the relation of ideas in physics to objective reality. Our ideas are never complete, but neither need they be totally wrong – even when they’re wrong. What I mean by the last statement is that my mental picture of how things work is resolutely Newtonian. We know that this isn’t how the world really operates, but we know that we get accurate answers in many circumstances in this way. So Newtonian physics is a “truth” about the universe, even if it’s not the full truth. I think of it as like projecting a shadow of some complex object: the shadow (Newtonian Physics here) doesn’t tell you what’s there in 3D, but it truly represents one aspect of that broader reality. I suppose this analogy can be continued by wondering whether a set of various different shadows (string theory?) can give you complete knowledge, to which one would suggest the counterexample of a sphere containing an internal cavity.

  8. telescoper says:

    Interesting, but string theory always suggests to me a large collection of spherical objects….

  9. andyxl says:

    John : most days my feeling is pretty much the same as yours – we are looking at shadows on Plato’s cave, but at least they are shadows of the real things, not just arbitrary operational sausage machines. (Boy is that a mixed metaphor. The shadow of an end-on sausage might like a tennis ball. How misleading is that ?)

    But I was hoping you would splutter and go for the “Music not very human” bit.

  10. Andy, I’m glad you liked the musical Turing test! I wonder if you’d still think that music is less human than conversation if a human playing human music were pitted against a computer playing computer music. I wonder – would this be testing for consciousness or something else?

    This reminds me of my favourite question for online Turing tests. Jim digs a cubic-metre-sized hole in an hour, and so does the big yellow digger. Does this prove that the big yellow digger has muscles? Of course not. So my question is: why do I think the Turing test is not a test of whether machines are conscious?

    Perhaps my analogy-spotting test is too stringent, because some astronomers who I won’t name failed this test. Hmm.

  11. John Peacock says:

    Andy: The “music not very human” thing didn’t worry me too much. But that’s because I always came at the Turing test in general from the opposite extreme that Alan Heavens (for example) seems to adopt. I can’t really think of any way that I could convince myself that I’m not just some elaborate equivalent of a computer responding to external stimuli according to hidden and complicated algorithms; so extending this to someone in another room seems wholly unnecessary. I think it’s obvious that we are all just water-based computers, so why worry that silicon computers can compose?

    In fact, I tried the musical Turing test, but got it wrong. The problem is that some of the computer compositions are closely based on an existing well-known piece; I couldn’t decide whether such parody was more likely to come from a human, or an over-constrained piece of code. I guessed the former, but it was the latter.

    This reminds me of Markov Chain algorithms for computer text generation (google markov text for free code). You supply a page or so of source text and the code makes a frequency table of N-tuples of consecutive words (pairs in the simplest N=2 case). You pick a starting word at random, and then the frequency table tells you what might follow, with what probability, and you draw a random number to decide which alternative to pick; and so on. The amazing thing is how plausible the N=2 realizations are, given that they know nothing about grammar etc. I once had to write some tedious report and was seriously fed up after 2-3 pages, so I Markoved it up to 50 pages and took it to coffee; everyone flicked through and said they thought it looked great. Now, the interesting thing is that even N=3 is much less interesting – you tend to get large blocks of verbatim quoting from the original, with just a few sideslips. But N=2 comes up with things that sound like the utterances of Spike Milligan: you keep thinking it’s about to make sense, but it never quite does.

  12. […] here:  Uncomputable Insight Category: Object | Tags: eagerly-offer, emperor, higher-art, metal-bach, most-people, […]

  13. Martin E. says:

    re: Turing tests, The Economist had a little piece called

    Polyphemus does the hoovering
    last week (23April2009), which gave me the willies. We’re used to computers taking snaps and being next to useless at decoding them – they *only* see the pixels, perfect Zen machines! But it turns out that moving the camera around to get multiple views makes it much easier to figure out what an object is*.
    So these moving robots construct a 3D model of the world, much as any animal must to move around. That is some small but significant step toward having a mind. At the same time facial recognition is coming along fast, so it may soon be possible to have a computer that can recognize an individual, and tell from that person’s expression how it feels, can follow the person’s gaze or pointing. In other words it could behave like a dog. In computer parlance, dogs have a ‘theory of mind’. That is they know that others have minds too. Suddenly, and a bit scarily, computers seem about to cross that threshold. Coo. Anyone else feeling a wee bit Luddite?

    *John: does this mean that if we have multiple theories, we see shadows of reality from many directions and so get a better picture of It All?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: