[From Hank Folson 970825]
(Bill Powers (970822.0632 MDT))
>>Rupert Young (970822.1300)--
>This is a nice example of the Test. On the surface, it looks like a
>simple open-loop sequence, but when it's disturbed, the wasp goes back to
>the point of the disturbance, corrects the error, and finishes out the
>sequence again, or tries to.
>The usual point of examples like this is to show how predetermined the
>behavior of the organism is. You were quite sharp to see that it actually
>shows a control system. The wasp is not controlling its behavior, but its
>perceptions; it changes its behavior to make the perceived sequence of
>consequences be right. If the behavior had really been pre-programmed,
>the wasp would have gone back to where the cricket _was_, and gone
>through the motions of dragging an imaginary cricket into the burrow.
How is this wasp example explained in terms of reinforcement? I am sure it
can be done. (see below)
ยทยทยท
------------------------------
Bill Powers (970822.0742 MDT)
>Browsing through odd files, I came across an essay by Avery Andrews on
>Speculations on Grammar, Semantics & PCT; Avery D Andrews; April 1996
>The assumption seems to be that if we could find a set of rules
>that correctly predicts how people talk, we would have explained how they
>talk. In the middle of rereading this essay, I realized that this
>assumption isn't necessarily true.
>Consider the following rules:
>If the pendulum is left of center and moving left, it decelerates.
>There is a similar set of rules that applies to the pendulum when it's
>right of center.
>The first question is, does this set of rules accurately predict what the
>pendulum will do? If I haven't made a mistake, the answer is yes. The
>second question is, do these rules explain the motions of the pendulum?
>The answer is no.
>The rules describe what happens but they don't
>determine or explain what happens.
>The basic question I'm raising is this: to what extent are rules of
>grammar or syntax mere descriptions, and to what extent are they
>explanations and determinants of linguistic forms? Obviously, rules which
>are only descriptive can't explain linguistic forms, and we have to
>explain the observations in terms that don't involve linguistic rules.
>
>Thanks to Dag Forssell for making this nice distinction between
>description and explanation in various of his essays.
These observations are about language, but don't they apply equally to
related areas such as the way researchers approach concepts like
'reinforcement'? They seem to me to apply in two ways: the use of language
in developing a theory, and the use of descriptive rules versus
explanatory rules.
Sincerely, Hank Folson