Re: Jeff Hawkins Q&A

On 4 Nov 2005 14:22:07 -0800, "feedbackdroids"
<feedbackdroids@xxxxxxxxx> in comp.ai.philosophy wrote:

>
>>
>> > OTOH, references to operation of
>> >the brain have always included the major elements found in systems
>> >covered by the term "mathematical complexity theory", namely, the many,
>> >the nonlinear, and the presence of feedback.
>>
>> Well from what I've seen of the behavior of mathematicians I doubt
>> they have any better grasp of the mechanics of linearity and non
>> linearity than the man on the moon. They are probably aware of many
>> features and aspects of complexity and non linearity but don't really
>> have a clue as to why these features are manifested and how.
>>
>
>
>No clue as to the "why" of it is probably valid when it comes to
>systems with many dimensions, such as the brain, but you also have to
>remember that complexity theory is only a couple of decades or so old,
>modern neuroscience not much more than that, and computer power is
>still marginal for solving really complex problems.

Computer power isn't the issue for nonlinear complexity.

>OTOH, it's not difficult to understand the workings of very simple
>systems that exhibit chaotic or self-organizing properties. By way of
>example, the classical population growth model, simple as it is,
>exhibits chaotic operation for large A ...
>
>dx / dt = A x (1-x)
>
>And if you look at it, you will notice it contains temporal dynamics
>[got to keep the temporal guys happy], as well as nonlinearities, as
>well as both positive and negative feedback terms. This equation
>produces choatic output, and is the simplest example there is.
>
>
>> >
>> >Don't know how many times it's been mentioned now, but systems in
>> >nature which exhibit what is termed complex organization and
>> >self-organization, all have (a) nonlinearities, and both (b) positive
>> >and (c) negative feedback, as "necessary" elements.
>>
>> Never suggested otherwise. All I claim is that we understand the
>> concepts of feedback and linear complexity quite well in mechanical
>> terms but not the concepts of nonlinearity and nonlinear complexity.
>> At least to judge by the behavior of mathematicians who can't even
>> seem to understand application of elementary infinitessimal integrals
>> for action as far as particle spin is concerned.
>
>
>Don't confuse physics / quantum theory with everything else.

I'm not really sure how quantum effects fit into intelligence theory
if at all.

>Personally, I think there must be a few empty holes there, waiting to
>be plugged.

Sure it would just be nice to get to them instead of moaning and
groaning over the numbers all the time. If nonlinear complexity
really represents the basic problem then the basic problem can't be
solved; it can only be addressed, recognized for what it is, and
described.

>> > Remove any of the
>> >3, and you get something with no more (self-) organizing capability
>> >than a random gas. That's the way it looks.
>>
>> That's the way it is. But linear complexity of itself has nothing to
>> do with intelligence which is the argument I thought you were making
>> with your references to numbers and combinations.
>
>
>I said the converse.

I'm not sure who said what at this point. What I remember people
saying was that the numbers and combinations are prohibitive but that
would be true whether the numbers themselves were prohibitive or not.
Nor do I see any evidence that anyone can say which combinations of
numbers are relevant. So citing combinations alone doesn't do much. It
might be that only a handful of combinations have any basic importance
to the artificial mechanization of intelligence. And that lumping them
all together as combinations would be nothing more than linear logic
applied to combinations as combinations instead of combinations as
intelligent combinations. If you're just going to look at combinations
as such the problem is self defeating to begin with.

>Which is exactly why
>> random gas clouds aren't intelligent regardless of the numbers. On the
>> other hand if you're talking nonlinear complexity, that emerges quite
>> early in number sequences such that between mechanical differences
>> between differences and non linear complexity I would imagine one
>> couldn't really make any sense of internal brain operations much over
>> twenty five or so differential elements.

>Obviously the problem isn't easy.

We're not talking about easy or hard for that matter. We're talking
about not understanding the problem of intelligence itself. People use
this difficulty to exculpate their inability to understand a problem
in mechanical terms. Once they understand it mechanically the problem
is easy whether or not it is easily implemented. Once understood
linear logic takes over. That's the sense in which I mean the problem
becomes easy.

>> However I'd like to stress once again that I'm not trying to put words
>> in your mouth. It's just that I've haven't been able to make much real
>> sense of your numerical and combinational complexity arguments so
>> I've resorted to a kind of revisionist interpretation for them. If
>> this offends you I'll be happy to omit references to you as far as my
>> own ideas on non linear logical and non linear complexity are
>> concerned. But I still believe the arguments offer considerable relief
>> from whatever ambiguity remains in terms of neural operations.
>>
>
>
>The short of it is that, if there is a magic bullet for solving the
>problem, it hasn't been devised yet.

You obviously can't know this at this point because you don't know
exactly what the problem of intelligence is. All we can really say at
this point is that there is no no linear logic applicable to the
mechaniczation of intelligence generally and no recognized solution
to the mechanization of intelligence for that reason. As soon as we
find the linear logic applicable to nonlinear logic and nonlinear
complexity we'll know what we're up against. But not until.

~v~~

.

Relevant Pages

  • Re: Jeff Hawkins Q&A
    ... >>>brain is responsible. ... Certainly Dan's arguments have all been about linear ... >> complexity, that is simple numerical terms and combinations. ... >the nonlinear, ...
    (comp.ai.philosophy)
  • Re: Jeff Hawkins Q&A
    ... >>brain is responsible. ... Certainly Dan's arguments have all been about linear ... > complexity, that is simple numerical terms and combinations. ... the nonlinear, ...
    (comp.ai.philosophy)
  • Re: Complexity science meets geology, some recent research papers
    ... Looking at the world through a non linear perspective takes ... It's also in how the word complexity is now defined. ... state is considered to be at criticality. ... For a cloud the opposite extremes are simply ...
    (sci.geo.geology)
  • Re: Any ideas how to linearize (and evaluate) this nonlinear equation?
    ... >nutrient, X0 is the total amount of the nutrient contained in all ... >point for the nonlinear one. ... if I had done a linear ... These four kinds of confidence intervals which you ...
    (sci.math.num-analysis)
  • Re: Jeff Hawkins Q&A
    ... > features and aspects of complexity and non linearity but don't really ... well as both positive and negative feedback terms. ... > concepts of feedback and linear complexity quite well in mechanical ... > terms but not the concepts of nonlinearity and nonlinear complexity. ...
    (comp.ai.philosophy)
Loading