Scientists 'locate' intelligence

Sergio Navega snavega at attglobal.net
Sat Aug 26 09:21:54 EST 2000


Glen M. Sizemore wrote in message
<6rCp5.2199$d8.954024 at typhoon.southeast.rr.com>...
>Glen M. Sizemore wrote in message ...
>>Sergio Navega:
>>>"...how to explain operant conditioning..."
>>
>>Bravo, sir, bravo!
>>Glen
>>
>
>Sergio: I'm doubtful about your intention here, if it is
>just an ironic comment or a sincere agreement with
>the difficulty of the task. If it is just an irony,
>I may even understand it, but only if you're confusing
>'operant conditioning' as being an explanation,
>instead of being what it is, just an evidence with
>a fancy name.
>
>Glen: It is my turn to be confused. Actually, my
>comment was not meant as "ironic"(I'm not sure
>what that would mean here - I think the English
>word you are looking for is "facetious") or as
>"...sincere agreement with the difficulty of the
>task..." (if the "task" is a neurobiological
>understanding of operant conditioning, even though
>you have my agreement here). My comment was
>meant as praise for what appeared to be your
>recognition of the overwhelming importance of the
>process (of operant conditioning) in understanding
>complex human behavior, but I see the praise was
>probably premature.
>

Our mutual misunderstanding can probably be traced
back to the inexpressive nature of textual
first contacts.

>It is true that the term "operant conditioning" is
>simply a name given to a set of observations, but
>there is a sense in which it is an explanatory
>mechanism. This is the same sense in which "natural
>selection" is an explanatory mechanism even if no
>reference is made to the "underlying mechanism."
>See? Or is it necessary for me to educate you?
>

Could you expand on what you think are the similarities
between explanatory nature of operant conditioning and
natural selection? The latter, for me, is supported by
a set of causally related concepts, while the former
(on my vision of what it means) is a name for what
happens in a specific situation, not a set of processes
causally related. I'm ready to be educated about that,
if you really have a distinction that goes beyond my
understanding.

>Sergio: Behaviorists frequently confuse both,
>thinking that there's nothing more to do after naming
>evidences and devising laws.
>
>Glen: At least you recognize the lawful nature of the
>experimental analysis of behavior (EAB), but most
>behaviorists that I know do not claim that operant
>conditioning cannot be understood in a
>neurobiological sense.

I don't want to be seen as if I am a behaviorist (I guess
you don't mind being taken by being one), but I really think
there are good things discovered by pure analysis of
behavior. And while I agree with you that there are
strong neurobiological support, I still miss important
parts of the question.  Donald Hebb, although he was
not a behaviorist himself, may be seen as giving important
insights to the nature of conditioning in a neural
level, things that are in constant confirmation by
recent neuroscientific research.

>Nor do they question the
>importance of this endeavor. They do, however,
>usually claim that behavior qua behavior can be
>understood at its own level. You seem to recognize
>as much when you talk about "devising laws."
>Perhaps it is you who are confused?
>


Not in that particular issue. As I said, I find
insufficient to devise laws. Without going to the level
of explanations, we will be stuck in predicting things
(which is already a damn useful thing), but that alone
does not allow us to do *create* things based on that
knowledge. To create things we must understand, we must
have explanations and models.

>Sergio: But science only grows if its
>laws go beyond mere predictive power (after all, we
>can predict very well what happens because of
>gravity; however we don't have a damn clue about
>how to explain it).
>
>Glen: Your argument is internally inconsistent. The
>EAB does, in fact, generate laws, and laws do more
>than predict - they allow one to control one's subject
>matter.

I have nothing against this. But I question that control
is everything. See, I'm from the AI side of the story.
Here we are not concerned in just finding laws of thought,
we're concerned in understanding the essential principles
so we can rebuild functionally similar mechanisms with
other technologies. That's (IMHO) the only way to go.
It is an essentially creative task, something that
involves understanding causal principles and structures,
devising theories and models, which is more than mere laws.
We must postulate productive, coherent and wide theories.

>
>Sergio: It's necessary to develop explanations which
>are reasonably intertwined and coherent with other
>theories, from other fields.
>
>Glen: This is naive at best. No one would try to
>reconcile, for example, special relativity with notions
>of a mechanical ether. Some theories are simply best
>abandoned. Some are not.
>

There's no doubt about that. The reconciling I was talking
about is among *equally interesting* viewpoints. I'm sure
that you'll take this with a grain of salt, but I want
to understand the issue so that I can explain equally well
what behaviorism and gestalt (among others) propose, at the
same time.

Why am I concerned with that? Because there are things in
gestaltianism that are *very interesting* and at
the same time don't fit well in behaviorism (the opposite
is also true). I don't want to be seduced by one strategy
at the cost of *hiding* the good things of other strategies.

In summary, it is not a question of refining behaviorism
or improving Gestalt, it is a question of understanding
better what nature is showing us.

>Sergio: But I guess you should've noticed that what
>I really proposed was a bit different:
>
>"...how to explain operant conditioning *and*
>systematicity/generativity simultaneously..."
>
>Glen: I have followed some of your posts and I find
>them occasionally ambiguous, as I find the statement
>above ambiguous. I am sorry that I have not
>"studied you" but I am waiting for a reason to do so.

Thank you very much for bringing my ambiguity to the
issue. Believe me, this is something that irks some people
here. Sometimes I'm accused of being dubious and vague.
Well, I can find a place for two kinds of people: those who
insist in seeing everything always in crystal clear ways
(even if this means rejecting some ideas) and those
who recognize that most things have elusive nature.

Things are not cristal clear. Things are ambiguous *by their
own nature*. If I sound ambiguous sometimes, it's because
the things I'm talking about are ambiguous in their
own natures (at least the way I see them). I think it
is a deep hindrance to our understanding to try to *always*
see things in a straight, clear way (although the opposite
is *also* bad). That may conduct us to "forget" things that
may be of crucial importance for the best way to see nature.
We must often review our premises, come back to where we
came from to see if we got the right path, without any
prejudices.

Now about my phrase above, conditioning is fine to explain
certain things, but what about the other side? What about
the kind of generative abilities that a child possesses?
It is not enough to say that it is a "random" combination
of previous conditionings to assemble something new. One
must explain down to the informational level what happens
there. Gestalt psychology gives us a lot of "insights"
(being a bit self-referential) about this topic and
surprisingly there are recent neurobiological work that
emphasizes this gestaltic nature of perception.

Remember that I said that behaviorism is getting support from
neuroscience. Now Gestalt also receives that kind of support.
How can that happen? What am I supposed to do? Forget
gestaltianism and stick with behaviorism? Forget behaviorism
and stick only with gestaltianism? I'm not satisfied with
both of them! I want something that can explain things better
than both strategies.

>The EAB is quite interested in a systematization of
>behavioral phenomena - it has been, historically,
>interested in an empirical systematization of
>behavior, and many have recently become interested
>in more mathematically oriented systematizations. I
>do not know what you mean by "generativity,"
>though.......perhaps "generality?" Again, my
>apologies if I have not studied all of your posts but
>you have given me little reason to do so, and your
>arrogance, and inclination to slander those you
>clearly do not understand in any scholarly way,
>seems to support this perception.

It is a wrong perception of yours, I assure you. I'm one
of the best "listeners" of this group. The only instances
of 'arrogance' that I may be accused of may be related to
responses to unusually aggressive comments of some special
posters here (one day perhaps you'll know this by yourself).

Anyway, I used "generativity" in a chomskyan sense, the
ability humans have of perceiving rule-like structures in
(for instance) language and then using these rules to
generate an unlimited number of sentences. Don't tell me
that behaviorists explain the creativity of language in
a satisfactory way, I'm not convinced (maybe you can
change my mind...nah!)

>
>Sergio: This is the kind of question that seems
>interesting to investigate on a wider approach than
>the purely behavioristic methodology.
>
>Glen: Behaviorism is not a methodology but, rather,
>a philosophy that holds that behavior qua behavior is
>the proper subject matter of psychology. It seeks to
>explain complex behavior, at the level of behavior, in
>terms of the facts uncovered by an experimental
>analysis, which emphasizes direct demonstrations of
>experimental control of the behavior of individual
>subjects.

I am in almost full agreement with what you say here.
I just think that there are things that must be posed
in terms of models ("mental models", forbidden word!).
Far from being a methodological sin, this is pretty
much the daily activity of most scientists such as
physicists, chemists, biologists, etc. They are
posing models and working with predictions, coherence
and falsification of theories.

>It is, thus, true that behaviorism is
>associated with a particular kind of methodology,
>but it is so much more, as is evident in such works
>as Science and Human Behavior (Skinner, 1957) and
>Verbal Behavior (Skinner, 1957).
>

One of my interesting readings is 'Understanding
behaviorism', William M. Baum. I belong to a strange
category, one that does not reject behaviorism but
also does not accept it as being the best (neither
unique) methodology to understand animal intelligence.
I keep asking questions that behaviorism (to my knowledge)
don't answer satisfactorily:

   Why the hell did Thorndike cat pulled the string?
   Why random behavior was selected by evolution?
   What's the relation between neural synchrony and conditioning?
   Why neural representation changes with different conditioning stimuli?
   ....... (two dozen more)

Although I'm still building an explanatory construct to deal
with all that, I can't think of doing this without resorting to
several different disciplines, without prejudices.

>Most cordially and respectfully,
>Glen
>


I thank you very much for disagreeing with me in a so polite
manner. For me this is the best way to use newsgroups, it is
the reason I keep insisting in coming here now and then.

Best Regards,
Sergio Navega.










More information about the Neur-sci mailing list