Generative vs. Explanatory

[from Gary Cziko 930110.2106]

I've been enjoying the discussion between Greg Williams, Bill Powers, and
Rick Marken, apparently originally sparked by my 3 Pencils adaptation of
Rick's research.

I certainly don't have time to get heavily involved in this, but I want to
make a quick comment on Greg Williams's (930109) saying:

f is supposed to USE observables only, like cursor position and handle
velocity. Of course, f itself is a model, and I suppose you could then argue
that behaviorists actually propose generative models if they do any more than
curve-fitting (per the above, the Newtons of their day!?!?). Bill argues below
that behaviorists do NOT propose generative models, but I suppose that he is
claiming that all they ever do is curve-fit. And here I am trying to claim
that at least some behaviorists go beyond curve-fitting (like Newton did) and
yet use only observables in their "models" -- which aren't the sorts of models
I've always been thinking underlying generative models to be. I think curve-
fitting doesn't count as making generative models, but I'm not convinced that
hypothesizing functions containing only observables does count as making
underlying generative models.

Perhaps we need to make a distinction between generative and explanatory.
Curve fitting can never be based solely on the observed data as there are
an infinite number of curves that can fit any collection of data points.
The functional relationships obtained by curve must therefore always go
beyond the data and the resulting hypothesized relatioship can also be used
to make predictions which have not yet been observed. Presumably, more
data will help to fine-tune the hypothesized relationship by showing that
it is in some ways invaled and new relationship is then hypothesized which
did everything the old could do and some things it couldn't.

But such a hypothesized relationship, although generative, doesn't explain
WHY the relationship exists. To explain we would have to move to an
underlying level, from psychology to physiology, for instance.

So perhaps a difference between behaviorism and PCT is that while both try
to be generative, only PCT tries to be explanatory. To make predictions we
need a generative theory. To answer "why" something happens, we need an
explanatory one.

It also appears to me that Chomsky's grammar is also generative but not
explanatory since he uses no physiology in his theorizing. Would Avery and
Bruce agree with this?

Where does this leave Newton?--Gary


Gary A. Cziko Telephone: 217-333-8527
Educational Psychology FAX: 217-244-7620
University of Illinois E-mail:
1310 S. Sixth Street Radio: N9MJZ
210 Education Building
Champaign, Illinois 61820-6990

[Avery Andrews 930111.0904]

(Gary Cziko 930110.2106)

>It also appears to me that Chomsky's grammar is also generative but not
>explanatory since he uses no physiology in his theorizing. Would Avery and
>Bruce agree with this?

Does this mean that Newton was non-explanatory? If so, yes. But I think
this goes beyond what many people mean by `explanatory'. What Newton did
to an extreme degree, and what Chomskyans do to an immensely lesser degree,
is reduce many apparently disparate facts to fewer common principles,
& this is all that many people mean by `explanation'.