metaphors

[From Bill Powers (951024.1010 MDT)]

francisco arocha( 95/10/24 08.51 EST) --
Joel Judd (951024.0750 CST) --

It's very nice to have some additional people (and I'm also thinking of
Shannon Williams as well as the standard bunch) arguing on my side. All
well said.

···

-----------------------------------------------------------------------
Bruce Abbott (951024.1000 EST) --

     Well, metaphorical or no, it [imprinting] does mean "leaving a mark
     on something." The "something" is the newly hatched bird and the
     "imprint" is a complex pattern of inputs that allow the bird to
     recognize its mother.

But don't you see that behind the way you say this there is a whole
theory of how stimuli affect organisms and how behavior works? You are
asserting that there is a complex pattern of inputs in the natural world
which, because of its organization, does something to the chick's brain
to make it behave in a particular way relative to its mother. You aren't
SHOWING that this is true, you're ASSERTING it, by all the words you
choose, before the argument can even begin. You're trying to get
agreement to all these subtle postulates by brute force, by subterfuge,
by allegation and implication and association, by any means but
explicitly proposing your model and defending it with reasoning and
facts. And what you're asserting is contrary to the general ideas of
PCT.

We do NOT observe any "mark" being made on the chick, on its outside or
in its brain. We do not observe the same patterns that the chick
observes. We do not, in fact, have the least idea how the chick
recognizes its mother, or even if "its mother" is what it recognizes, or
if "recognizing" is the right word for the process. We don't know what
goals the chick inherits, or how it manages to satisfy them.

     Imprinting is the process by which this unique pattern comes to be
     stored in the chick's brain, as evidenced by the chick's ability
     thereafter to respond uniquely to its own mother as opposed to
     other female hens.

We don't know that any unique pattern comes to be stored in the chick's
brain. We don't know that the chick's action is a "response" rather than
a means of controlling a perception. In fact, the current PCT assumption
about memory is that only the values of signals are stored; the pattern
is recreated by the perceptual functions when the signal values are
replayed into them. Perceptrons do not work by storing patterns, either.
You're asserting a particular theory of perception. We don't know that
the chick is "responding uniquely" to "its mother." And you haven't
described any process; you've only claimed that there is one, and that
it's to be called (for some unfathomable reason) "imprinting."

This whole way of framing the observations shows only that you are up to
your eyeballs in a particular theoretical frame of reference which is so
ingrained that you don't even recognize theoretical statements when you
are making them. Everything you are saying sounds completely reasonable
to you, I'm sure, and I'm also sure that you can't understand why most
of the PCTers on this net find your descriptions doggedly biased in
favor of a non-PCT world view. Speak up, PCTers, am I right about that?

     We have no words to describe what changes take place in the brain
     that allow the chick to recognize its mother, once it has
     identified her.

But you have just used a whole string of words that DOES offer a
description: "changes in the brain" that allow "the chick" to
"recognize" "it's mother" once it has "identified" her. Even to talk
about the chick identifying "its mother" implies that the relation of
mother to chick is as we human beings see it (by analogy to the
relationship between mother and child), and that all that is involved is
"identifying" something that already exists. The whole picture is that
of the environment doing something to the chick's brain, instead of the
chick's brain making whatever sense of its raw intensity inputs that it
can, given its capacity to learn and the influence of inherited ways of
perceiving. Each time you use language like this you are denying the PCT
approach, which is centered in the behaving organism, not in its
environment or in the observer's own informal habits of perception.

The whole point of PCT methodology is to substitute formal tests on
which we can agree for the individual free associations that
undisciplined or egocentric observers bring to the study of behavior.
The less we rely on informal language and metaphor, the more nearly we
can achieve the scientific ideal of public observations interpreted and
analyzed by public means according to an explicitly-stated framework
that anyone can apply to get the same results and conclusions.

     Almost all nouns are metaphores. Take the term "signal," as in
     "reference signal." According to my dictionary, it comes from the
     latin for "sign"; by claiming that a control system contains a
     reference signal, you are literally saying that it contains a sign
     -- a conventional symbol representing an idea.

Yes, I'd accept that as one valid way of describing a reference signal.
I don't know about the term "contains," which implies that the sign is
inside the reference signal; I'd rather say that the reference signal IS
the sign, the symbol. What else would a sign or symbol be? Of course we
have no way of knowing what the Romans meant when they said "signum."

     "Reference" comes from the latin meaning "to bear." Your phrase
     "reference signal" therefore literally means "that which bears a
     sign." I don't know why you choose such metaphores to convey your
     ideas; PCT researchers will be looking all over the brain for
     structures bearing signs. (;->

It means to bear again or back, if you want to get picky. But few people
nowadays would make that association. It's not hard to explain which of
the several common meanings is intended in PCT; we don't mean a letter
stating one's qualifications, or a citation of someone's paper, or the
implication of a remark, and so forth. The common meaning that is
intended is that of "reference mark" or "reference scale," something
against which something else is compared. If you don't like "reference"
there are many other words one could use instead: target, standard,
goal, intention, and so forth. A person who understands PCT could
explain it without using any of the standard terms.

     At this point I hope you are saying to yourself, "why, that's
     absurd!" I hope so, because there is really no difference between
     words like "control" and "imprinting," so what applies to one
     applies to them all.

That's just not true. What most people mean by control is what we in PCT
mean by control. But what most people mean by imprinting is not what
Lorenz meant by it -- I hope. If Lorenz did intend any of the common
meanings, then he had a very peculiar concept of how a brain works. Does
a perception really make a little dent in the brain?

     "Control" comes from the latin "contre" + "role," meaning role or
     catalogue, and originally applied to the checking or regulating of
     payments. Only by analogy does it come to mean "to keep within
     limits" such variables as speed or intensity. "Control" is a
     metaphore.

But in the original usage it was not a metaphor. Converted to the
appropriate part of speech, counter-rolling would mean seeing to it that
one way of checking the books agreed with another, which is certainly a
control process as we now use the word. The form is not that of a
metaphor, but of synechdoche, referring to a whole in terms of one of
its parts or vice versa, as in "the cowboy counted 20 head," referring
to the "controls" of an airplane, or reporting how many ships are
approaching as "10 sail." A counter-roll was part of a control process.

     The term "imprinting" simply calls attention to a specific kind of
     change observed to occur under a specific circumstance. The
     observable change in behavior leads inescapably to the inference
     that some rather permanent change has been laid down somehow,
     somewhere, in the imprinted bird's brain.

That is a _theory_ about what has happened, not an "inescapable
inference." You make the inference that supports your world-view; all
that makes it inescapable is that you can't think of a different one.
What if the bird makes that change in its own organization, because of
an inner inherited demand that some large object be maintained present
in the visual field? That's another "inescapable inference" according to
a different world-view.

     There is no appeal to magic. Proposing mechanisms and doing the
     research to test those proposals is what comes next, after the
     phenomenon has been identified and labeled.

Come on, the very way you describe the phenomenon is a proposal of a
mechanism. The language in which you are describing the phenomenon is
simply not objective: it contains a theoretical spin from the first
word. You speak of the "imprinted bird" as if something has done
something to the bird -- that's not "proposing a mechanism?"
-----------------------------------------------------------------------
An aside, to Gary Cziko or anyone else with a literary acquiantance with
the French language. I'm reading Montainge in translation (filling in a
tiny bit of my deficient classical education), and came across this in
_Montaigne: Essays_, Book 1, Chapter 7, first page, titled "That our
actions should be judged by our intentions":

     We cannot be held responsible beyond our strength and means, since
     the resulting events are quite outside our control and, in fact, we
     have power over nothing but our will; which is the basis on which
     all rules concerning man's duty must of necessity be founded.

Aside from the relevance to PCT, this passage contains the word
"control" supplied by the translator, in a sense that fits with PCT. The
first two books of Montaigne's Essays were published in 1580, so it
would be interesting to know what the original word was that was
translated as "control." Somehow I doubt that it was "counter-roll". Any
help here from someone with access to an academic library and able to
read the original French?
-----------------------------------------------------------------------
Best to all,

Bill P.