Newsgroups: comp.ai.philosophy
From: lupton@luptonpj.demon.co.uk (Peter Lupton)
Path: cantaloupe.srv.cs.cmu.edu!das-news.harvard.edu!news2.near.net!MathWorks.Com!europa.eng.gtefsd.com!howland.reston.ans.net!news.sprintlink.net!demon!luptonpj.demon.co.uk!lupton
Subject: Re: Is Common Sense Explicit or Implicit?
References: <35qk9k$6tb@mp.cs.niu.edu> <35ak50$d3v@mp.cs.niu.edu> <357pp8$p9i@mp.cs.niu.edu> <357uj6$m8r@search01.news.aol.com> <358haj$ali@mp.cs.niu.edu> <359934$jrp@agate.berkeley.edu> <359cus$k0s@mp.cs.niu.edu> <35af8j$4r9@agate.berkeley.edu> <227798242wnr@luptonpj.demon.co.uk>
Distribution: world
Organization: No Organisation
Reply-To: lupton@luptonpj.demon.co.uk
X-Newsreader: Newswin Alpha 0.4
Lines:  261
Date: Fri, 23 Sep 1994 22:37:21 +0000
Message-ID: <48020715wnr@luptonpj.demon.co.uk>
Sender: usenet@demon.co.uk

 
> >> Yet, although they are not linguistic, there is much in 
> >> the behavior of apes to suggest that their cognition is 
> >> very like ours. That suggests to me that evolution does 
> >> not need beliefs.  
> 
> >Evolution of what? Clearly *evolution* does not need beliefs.
> 
> Alright, I chose my words carelessly.  My point was that evolution
> can lead to complex organisms such as apes, without the apparent
> need for beliefs.

Agreed - so long as you concentrate on those things one can do without
beliefs, you will readily be able to show that those very things can,
indeed, be done without beliefs.  

> >> Thus I see the existence of beliefs
> >> more as an incidental side effect than as an important cause.
> 
> >Cause of what? Of what it is to be human rather than ape? I think not.
> 
> That is a misinterpretation.  I was referring to the supposed
> importance of beliefs as causes of behavior.

It seems to me that there are very real and very great differences between
what apes do and what humans do. Or am I missing something?

> >> In order to understand the cognitive development of homo sapiens from
> >> earlier apes, we need to think about what is required for early
> >> hunter-gatherer man.  And it seems to me that deductive inferences
> >> from beliefs could not have been of central importance.
> 
> >Animals certainly have many cognitive abilities that we share. 
> >I will also agree that our linguistic and propositional abilities are
> >grounded in abilities which cannot themselves be expressed in entirely
> >propositional terms.
> 
> Ok.  The need for abilities not easily expressed in propositional
> form is what I was arguing for.

This above is really quite innoccuous - I suspect that very few people
would disagree with you here. What is contentious in what you say is the
attempt to down-play the role of linguistic and propositional abilities.
Why do you want to do this? What is the motive? These questions are
not rhetorical, they stem from a genuine puzzlement. 

I would go further than the above. I would say that non-propositional
abilities (pre-propositional?) are prior to propositional 
abilities in the sense that propositional abilities are parasitic 
upon pre-propositional abilities: animals have pre-popositional 
abilities but not the propositional sort but you could not have 
propositional abilities without the pre-propositional sort.

The interesting thing, from my point of view, is that our propositional
abilities "grade off" into pre-propositional abilities - there is
no sharp line. Rather there is an increasing uneasiness in marginal
cases that true/false are out of place. (Marginal not in the sense
of unimportant but in the sense of not knowing whether something
is true or false). On the other hand, it is all too easy to
think of cases where true/false seem as right as right can be.

> >Indeed, I have argued that simplificational abilities are exactly
> >of that non propositional sort.
> 
> I do not have any serious disagreements with this, although I am not
> convinced that 'simplification' is the correct terminology for what
> is needed.  When you discussed your ideas of simplification some
> months ago, my disagreement was not with the basic idea, but with the
> way you wanted to define and implement simplification.

I am still talking about Algorithmic Complexity.
However, I have learned, since then, that the objectivism/subjectivism
debate is not for me - I shall avoid it as much as possible for the simple
reason that I can see no benefit in continuing it. This is not intended
as a slight to anyone, just a realisation that it is an unproductive
activity.

> >Although animals do not have language
> >and propositions, they do learn and this involves trial and error. 
> 
> Although I agree that animals use trial and error, it would be hasty
> to jump to the conclusion that learning necessarily requires trial
> and error.  

Fair enough. What I really wanted to say is that learning is the
acquisition of knowledge. Now the way you use the word "knowledge" casts
the net rather wide - far wider than beliefs, for example. This is very
much to my taste, also. So what I want to do is to point out that 
knowledge, widely drawn, includes far more than propositions and their
binary true/false status. Even so, there is still a notion of error
lurking there which, I think, can be a matter of degree and gives rise
to the notion that the animal might undergo revision and correction, say.

I think these words (revision, correction and error) do, indeed, 
belong with knowledge. If you don't think they do, that is fine, 
but I guess I would need an example of the sort of case you have in 
mind. I would need to know what the knowledge consisted in; how the 
knowledge was to be used; why the notions of revision and correction 
were inapplicable, say. In short I would need to know how your notion of
knowledge becomes wider than the scope of revision, correction and error.

> Of course, in any particular instance it might always be
> possible to apply a subjective interpretation, and thereby interpret
> some of the animal's behavior as consisting of trial and error.

I am not really interested in this, since, presumably, it is always
possible to interpret anything as anything else, given enough
subjectivity. (You can buy subjectivity at street corners, I believe).

> >That is, there is a notion of error which can be properly attributed 
> >to animals which learn. This, I claim, is accounted for by the 
> >disposition of animals (and not rocks) to simplify sensory data. 
> 
> Rocks surely simplify their sensory data.  Their thermal properties
> cause them to average their temperature over a period of time, and
> this averaged curve is simpler than the received thermal signal.
> Perhaps the previous sentence illustrates why I think "simplify" may
> be the wrong term.

Rocks don't have sensory data and they don't simplify. Apart from
an idiosyncratic use of these terms ("sensory data" and "simplify")
I don't see what you could mean.

> >Although simplification is not propositional, there is a clear
> >relationship between simplicity and the notion of fit we associate
> >with propositional content.
> 
> >Although non-human animals have many cognitive abilities these are,
> >in the main, limited to the here-and-now.
> 
> Nest building by birds; dam construction by beavers; the collection
> of reserves of nectar by honey bees - it seems to me that there is
> much in animal behavior which is not limited to the here and now.  We
> may argue that it is usually genetically determined, rather than
> learned, but it is still a form of planning behavior.

Sure. That is the sort of case I had in mind when I said "in the main".
My point is this. Certainly animals show the ability to learn which
involves the ability to detect and project regularities from one time
and place to a subsequent time and place. But the activation of such 
knowledge seems to be brought about by the situation, by the context, 
by the circumstances of use.

> >In order to release those
> >abilities from such constraints, simplificational abilities must
> >be instrumented in some way.
> 
> Perhaps you could explain your thinking behind this with a little
> more detail.  In particular I am interested in knowing why it *must*
> be instrumented, and I could use a clarification of exactly what you
> mean by "instrumented."

What I mean is this. A simplificational structure is a complex entity
which, in various specific contexts, will permit the organism to make
all sorts of interesting and pertinent projections of regularities.
There is no reason for that structure (and hence the knowledge embodied in 
that structure) to be driven solely by sensory inputs. Why not
drive the structure with "as if", hypothetical inputs? And why should
those inputs be at the level of sensory data? Why not use the 
classifications which the simplificational structure has already 
identified? Doing this (both in driving the structure and in getting
results) I have called "intrumentation" - it is an engineering term
whereby a model (or the actual artifact) is "instrumented" in 
order to record and measure experimental results.

For an engineering model, instrumentation consists in a few dials
and displays (actually, it can be very complex indeed, but that is
another matter). For a simplificational structure, there is not, in any 
interesting sense, a readily identifiable finite set of dials to set,
meters to read. Instead, one must be able to generate hypothetical
situations and sorts of situations on the fly - this begins to look
very much like language.

> >                             This instrumentation permits humans 
> >to consider possibilities, to organise their future actions in a 
> >way that non-human animals can do in only a very limited way.
> 
> >> ...it seems to me that deductive inferences
> >> from beliefs could not have been of central importance.
> 
> >What is lost by removing deductive inferences? It depends,
> >of course, on how narrowly or how widely the notion of 
> >deductive inference is drawn. Beliefs are to do with, not 
> >deduction per se, but with symbolic forms capable of 
> >being manipulated, combined, modified in a rational way 
> >even when that manipulation is far away from the time and 
> >place at which the underlying knowledge (fixed in the 
> >simplificational substrate) was initially gained.
> 
> Unless you have a non-standard meaning for "rational", it would seem
> that you have implied deduction in a sentence purporting to show the
> value of belief apart from the use of deduction.

That wasn't my intention. My intention was to focus in, not on the
word "belief" (which I don't think is really helping anyone) but
on the ability that those who argue for beliefs really want to
hang on to (including, of course, myself). First find the value,
then work out what sort of a thing or dynamical system gives rise
to that value. What I was arguing for was the value of being able
to lift our abilities from the here and now. There is a quite
separate question of what sort of entity or dynamical structure
is required to bring this about. Then there is a further question
of what the relationship is between such a structure and folk
psychology.

The word "rational" was introduced to point out that the sort of
inference we value is very definitely not *logic* as such. (I tend
to think "logic" when I hear "deduction"). The sort of inference
we often do is, strictly speaking, invalid and - should one ever
make such deductions part of a system of logic - would either form 
an inconsistent system or be extraordinarily weak.

> It seems to me that learning, which to some extent is present in all
> mammalian species, already allows knowledge fixed in the
> simplificational substrate to be used far away from the time and
> place at which that knowledge was acquired.  

Yes, we certainly agree here. But that is not the ability I was pointing
to. The words I used were: 

         "symbolic forms capable of  
          being manipulated, combined, modified in a rational way 
          even when that manipulation is far away from the time and 
          place at which the underlying knowledge (fixed in the 
          simplificational substrate) was initially gained"

It is this ability to *manipulate, combine, modify* in a rational
way which is so characteristic of humans. It is a quite unique
ability and permits humans to *invest* in these abilities independently
of actual use. (There will, of course, be instances of use, but the
investment in simplification - reasoning, thinking - can be quite
distinct).

> But, as far as I can
> tell, such learning need not require anything which fits standard
> interpretations of "rational".

Of course, if you restrict yourself to "such learning"! :-)

> >Now 
> >it seems to me that this ability - the ability to piece 
> >together bits and pieces learned at various times and 
> >various places to make a coordinated whole, a plan of 
> >action - would be of very real importance to early humans.
> 
> It is my impression that long range planning was not nearly as
> important to early hunter-gatherer societies as it is to modern
> civilized societies.

I was not talking about "long-range planning" exclusively. The plan 
could be carried out right away. The crucial point is that the mental 
resources being called upon in order to construct the plan and to 
organise one's future actions (not necessarily the long-range future)
are not driven by what is in front of the human here-and-now. 
The mental resources the human can call upon are whatever resources
have been instrumented, whereas the animal is restricted to those
resources activated by the here-and-now - that's the difference.

-------------------
Peter Lupton
