[Wittrs] Is Homeostasis the Answer? (Re: Variations in the Idea of Consciousness)
- From: "SWM" <SWMirsky@xxxxxxx>
- To: wittrsamr@xxxxxxxxxxxxx
- Date: Fri, 05 Feb 2010 14:07:07 -0000
--- In Wittrs@xxxxxxxxxxxxxxx, "iro3isdx" <xznwrjnk-evca@...> wrote:
> --- In Wittrs@xxxxxxxxxxxxxxx, "SWM" <SWMirsky@> wrote:
>
>
<snip>
>
> A homeostatic process is self aware (in a primitive sense) and is
> adaptive to change. Those seem like plausible precursors to
> consciousness and intelligence.
>
Yes, self-aware in a very primitive sense, i.e., in the sense that it is
equipped to maintain a degree of internal integrity or it could not be what it
is. So your point then is that it is from this limited sense of
"self-awareness" that our awareness of our selves as selves (our consciousness)
arises? I think that's a fair point and probably a useful observation because
it is certainly something about living organisms that prompts some to become
conscious over the course of their evolutionary history.
What I am interested to know, of course, is a little different, i.e., it's the
mechanism, or whatever one wishes to call it, that actually effects awareness
as we have it in ourselves and I don't think an evolutionary account (however
true and reasonable it is) does more than contribute to an understanding of
that.
>
> > Every time I ask you to answer these questions you tell me I am
> > misunderstanding you or talking about something else.
>
> I had been attempting to give explicit things that a cognitive agent
> does, but that AI people are not considering. And that's where
> communication breaks down.
>
> Let me try a different example. I take out a ruler and measure my desk
> to be 30 inches high. That "30 inches high" is a representation that I
> created out of whole cloth.
And which already requires a high level of conscious development in you, the
measurer. The question before us is where does that development come from,
what is there about you (or any of us) that makes us measurers in this way?
> By that, I mean that if I had looked at
> all of the signals being received by sensory cells, "30 inches high" is
> not something I could extract from those signals. It didn't come from
> signals, it came from my carrying out a procedure (a measuring
> procedure).
>
Yes, but we can't say that the phenomenon of being a consciousness with the
capacity to measure comes from having the capacity to measure, can we? The
latter depends on the former.
> The reason that "30 inches high" is about something, is that I created
> specifically to be about something. By constrast, the usual AI
> approach is to look at signals picked up by sensors. But those signals
> are not intentional.
By "intentional signals" then, you mean signals that we integrate into a
framework of meaning, signals that take on intention because they fit into our
existing structure of data association? And non-intentional signals are just
raw events, information for no one because no association is going on?
But don't you see that the very question is being missed in all this because
what really needs to be explained is how such signals do take on meaning,
intention, in the process of being received and stored by a conscious system.
The issue is WHAT is this associative process that links signals and thereby
gives them form and meaning?
The AI project seems to be grounded in the supposition that we can build
computationally based systems that do the same kinds of associative linking of
raw data as we do to build up pictures of the world which overlay one another
and cross link and which include both external aspects (our environment) and
internal (our bodies and memories) and which become the basis of our being
aware of things, of being intentional.
Now AI may, indeed, be the wrong way to go about this. It may be that, as
someone like Hawkins suggests, brains don't accomplish the task we are trying
to accomplish computationally (using complex algorithms running on computers).
But then, whether a computational approach could work, whether it could still
do the same thing as brains do, even if they are running on a different model,
would be a different question. That is, there could conceivably be more than
one way to achieve consciousness.
You have suggested that AI cannot work because it cannot be homeostatic. (Have
I got that right?) But aside from the question of whether virtual homeostasis
could do the trick or not (and we have seen that at least some AI people think
it could), perhaps we also need to consider whether it is homeostasis per se
that provides the requisite mechanism (in which case computers lacking this
mechanism might be expected to fail) or whether it just provides the impetus to
develop the mechanism (in which case the absence of homeostasis would not
necessarily mean an absense of the requisite mechanism -- and then there is no
reason to think AI must fail).
> They are just apparently meaningless signals
> picked up. There might be something useful hidden in them, if you have
> prior knowledge on how to extract. But how you get from signal to
> intentional representation is not obvious if it is even possible.
>
One way would be to suppose we consciously make sense of them, but that, of
course, begs the question. Another way is to suppose that we make sense of them
at multiple levels, only one (or a few) of which occur(s) at a level at which
we are aware, have access, and this(these) would be the conscious level. Thus
we could say that consciousness is part of a layered system of brain operations
that build on, and undergird, one another. But this IS consistent with the AI
thesis (see Minsky's The Emotion Machine). This is certainly consistent with
Dennett's model.
> So notice the difference. I start with intentions and deal with
> intentional representations from the get go. The AI model starts with
> meaningless signals, and I am inclined to think that it will never have
> more than meaningless signals.
>
But conscious intentions are not the same as non-conscious intention. The
lizard in the maze has more limited capacity to make sense of its surroundings
than the mouse. Yet both, as you would say, are homeostatic systems seeing to
maintain their internal integrity. Our intentionality in a maze would be
different again. As would the snail's and the earthworm's. I think we have to
be careful to separate intentionality as we find it in ourselves with lower
level intentionality which is, the deeper down you go, basically mindless
(unaware).
Insofar as it is missing genuine awareness, the intentionality of lower animals
starts to look a lot like what machines (in the sense of human artifacts built
to accomplish functions) can do. The question, then, is whether we can get our
kind of awareness (what we associate with being conscious) by building on a
platform of machine intentionality the way we get our kind of awareness on a
platform of lower level, unaware intentionality in the animal kingdom along the
evolutionary hierarchy.
> Think, for a moment about bird songs or whale sounds. It is quite
> likely that these are some part of a communication system. And if they
> are, then presumably the bird song is intentional for the birds. But we
> apparently cannot determine what those bird songs are about. At best we
> can look for correlations with behavior. If it were ever possible to
> start with signals that are meaningless to us, and to somehow find
> meaning by analysis of those signals, then this should be an ideal
> case. If we can't do it, then there's not much chance that we can
> program computers to do it.
>
Why do you think we can't do it? There has been lots of work in studying and
interpreting animal signals and lots of progress. Insofar as some signaling in
the animal kingdom may be fairly sophisticated (the whales' songs, dolphins'
chittering clicks, chimp screeching and gesturing) why would we think that
figuring meanings out here should be closed off to us?
And if we can crack the relevant codes (as we have done in some cases), why
wouldn't a computer be able to? But then the issue is not whether a computer,
as a tool for analyzing reams of data, could do it, but whether a computer
could be built that would have the same kinds of layered relational pictures of
the universe in which it exists that we seem to have and could then discern
significance in otherwise meaningless signals through use of such a network of
representations?
>
> > If your idea of "mechanism" is as constrained as you have described
> > it above, "human artifacts" only, then it would not be surprising
> > that you would make this mistake. But biological functioning is as
> > mechanical (on my broader view of "mechanism") as anything else,
> > even if they aren't manmade!
>
> If everything is a mechanism, the "mechanism" loses its meaning. I
> don't see anything mechanical about biological functioning. But
> obviously we disagree on what the word means.
>
>
Hmmm, that's misleading Neil. I never said everything is a mechanism. I
recognize, for instance, that we have lots of other things in the universe
including objects and games and activities and beliefs and so forth.
But I do think it makes perfect sense to speak of "mechanism" as being more
than just machines built by creatures like ourselves. Indeed, I think that our
language treats "mechanism" like that.
Here is one dictionary entry concerning the term:
http://www.merriam-webster.com/netdict/mechanism
Main Entry: mech·a·nism
Pronunciation: \ˈme-kə-ˌni-zəm\
Function: noun
Date: 1662
1 a : a piece of machinery b : a process, technique, or system for achieving a
result
2 : mechanical operation or action : working 2
3 : a doctrine that holds natural processes (as of life) to be mechanically
determined and capable of complete explanation by the laws of physics and
chemistry
4 : the fundamental processes involved in or responsible for an action,
reaction, or other natural phenomenon ? compare defense mechanism
Note that my use accords quite readily with entry #4 whereas you seem to want
to restrict our usage to #1a. But why should we do that? Why shouldn't we avail
ourselves of the full meaning of a term like "mechanism" in discussions like
this?
> > But you seem to want to disregard mechanical explanations entirely on
> > the grounds that this implies something artifactual made by humans.
>
> That's a complete misreading. I am not saying that we should discard
> mechanical explanations, and I am not saying that there's a problem
> with artifacts. I was just responding to your thinking it strange that
> there can be cognitive agents in a world of inanimate things.
When I made that statement I was speaking rhetorically, i.e., saying this is
why we are moved to wonder about where consciousness comes from because it
appears to be at odds with certain intuitions we have. It wasn't my claim that
I think it strange! It was my statement that we tend to think it strange when
we consider what minds seem, at first consideration, to be. Moreover, if I
recall rightly, I only made that rhetorical point after you had decried my
suggestion that we have to look for a mechanism underlying consciousness.
> My point
> was that, if anything, it is the presence of mechanisms that is
> strange.
>
> Regards,
> Neil
>
> =========================================
You mean human artifactual mechanisms or mechanisms as in "the fundamental
processes involved in or responsible for an action, reaction, or other natural
phenomenon"?
SWM
=========================================
Need Something? Check here: http://ludwig.squarespace.com/wittrslinks/
Other related posts:
