(penni sibun 920816)
[marken 920814]
I must also mention that Turvey has the honor of having published the
longest, most systematic and most incorrect critique of PCT to
date.
i'll believe you. i don't think that detracts from the fact that he's
not a behaviorist.
There are many incarnations of cognitivism. But what is common to all
of them (and what is commonly wrong) is that they view behavior as the
end result of mental processing. PCT says that it is perception that is
the result of "mental events" (reference signals).
i think this charaterization overstates cognitivism. i don't think
cognitivism necessarily sees beh. as the ``end result'' of mental
processing; i think cognitivism sees mental processing as crucial for
behavior.
This apparently small
distinction is as big as the distinction between behaviorism (which
sees external events as the cause of behavior) and PCT.
no, it's not: they both have the word ``mental'' in them, and
behaviorism clearly does not. the emphasis on the ``mental''
distinguishes cognitivism and pct from behaviorism.
There are some cognitive models that have some closed loop character-
istics; and probably qualify as control system models of the control
of higher level variables. An example is Simon's theor[e]m prover (and
current models that are similar, like the prolog language itself).
These models try to satisfy certain logical conditions (the reference
condition) by trying to satisfy other logical conditions in their
database. These are control models that work in a logical (not a
real) environment (the database). These are interesting models and
they may eventually become useful as models of what Powers calls
"program level" control. I have no beef with these models -- I think
they are very clever. They are just not yet connected in a realistic
way to behavior.
you don't find the fact that they are intractable problematic?
interesting (but somewhat off the point).
>if you think it's too weird to consider that organism and environment
>are fundamentally inseparable, why not consider a slightly different
>take on it: action and perception are inseparable; there is no way
>you can draw a line and say that on one side is the organism's action
>and on the other is the organism's perception. on this view, what
>could it mean to locate control in the head of the organism?
In Bill Powers' "Living control systems" page 251-252, precisely
this question is answered, simply and quantitatively. The short
answer is "because the organism amplifies, the environment dampens".
(my quotes, not Bill's).
this is simply sloganeering; it doesn't address the question in any
obvious way (though i realize it's been addressed in later messages).
>on the interactionist view, control is neither in the head not in the
>environment.
Sounds fine; but it turns out to be wrong -- a common problem when
trying to understand the world with words instead of models.
do you think you could cut down on the _ad hominem_ rhetoric?
I would like to learn about interactionism. But I
didn't really learn much from the quotes you provided.
fair enough. read almost anything by agre. i can give you preston's
address and you can ask for her paper.
Perhaps
you could describe an interactionist model of some very
simple phenomenon.
see below about driving.
Matter of fact, perhaps you might start by
explaining the nature of the phenomenon that interactionism
explains.
behavior. the same phenomenon that cogitivism and behaviorism
explains.
Does it explain the phenomenon of control? If so, how?
no. it does not seek to explain control. unlike cognitivism and
behaviorism, interactionism is not founded on the premise that the
idea of control is crucial to understanding or explaining behavior.
[from Avery Andrews (920815)]
(penni sibun 920814, on marken 920808)
What I'd say to this is that PCT is not cognitivist, because the *phenomenon*
of control is only found when an organism (or other `Agent') is put in an
appropriate environment (e.g. if the road is covered with an oil-slick,
one ceases to observe control of the direction in which the car is going).
The phenomenon of control is just what Rick says it is: when you introduce
disturbances that you would expect to change some aspect of the environment,
but that aspect doesn't change, because the Agent consistently does things
that `mysteriously' counteract your attempted disturbances.
on an interactionist view, there's nothing mysterious going on.
(though, frankly, i must confess that i don't really understand
y'all's rhetorical use of ``mysterious''; i don't really know what's
supposed to be mysterious to whom.)
PCT goes on to say (at least) two further things:
a) an Agent that is efficacious in the real world must effect control
b) the only way control can be effected is by means of certain kinds
of internal arrangments - perceptors, comparators & effectors
appropriately connected so as to constitute negative feedback loops
in the context of the actual environment.
i think it's conceivable that pct-type control might be more
interactionist than any of us understands at this point. however,
part b) is squarely cognitivist: it requires an inside-outside line
to be drawn, and puts crucial stuff on the inside of the line.
[From Bill Powers (920815.0800)]
Penni Sibun (920814) --
You say some things, and represent others as saying some things, that fall
strangely on my ear. For example:
>for these folks
>and others in ai, for gibsonian psychologists, for
>ethnomethodologists, etc., agents and their worlds are *not
>separable*. how can you have a respond to b's stimulus when a and b
>are the same thing?
If they can't separate agents and their worlds, they're trying to use the
same model for the inanimate environment and for living systems. This is
simply a mistake: the inanimate world does not run by the same rules as the
animate one. Of course if you jack yourself up to a high enough level of
abstraction, everything is the same as everything.
hold on. doesn't everything bottom out in physics? certainly the
current view is that everything is made of the same subatomic building
blocks, in a very fundamental, not abstract, way.
But then you have
nothing interesting to say, because all the interesting things we say about
ourselves and the world are about relationships between independent things.
If you can't tell things apart, because of having abstracted too far, then
you can't see any relationships.
ok, i agree that when we say interesting things, we say things about
objects and relationships that we have individuated for the purpose of
talking about them; certainly there are things that we typically
individuate (like organisms) so that we can talk to each other. on
the interactionist view, there is nothing sacrosanct about these
typical individuations. whenever you look very closely at a clearly
drawn line, you begin to see that it gets rather fuzzy. the line
between action and perception is a paradigmatic example of this.
It's hard for me to see how
a person who doesn't make this distinction [bet. perception and
action] can understand what it means to
say that action controls perception, unless that person means that
perception IS action, or that perception follows faithfully from action
(which it doesn't -- that's not the meaning of "control").
all true if you assume control. but as i said, the interactionist
view doesn't assume a model of control.
But action and perception are completely separable; among our most basic
and elementary demonstrations of control processes is one that shows an
almost total LACK of correlation between a person's actions and the
perception that they control. If you don't understand this fundamental
separation, then you've missed a basic point about how control works.
This is such a basic concept that I don't think we can understand each
other until we reach some sort of agreement on it.
i understand what you're saying, and i do in fact agree that you have
an interesting and efficacious model of control. you can build little
creatures who do cool things with your model. you might even be able
to build little creatures that are better than anyone else's (though i
think the jury's still out on that one).
but i don't think your model is sufficient to account for human
behavior (or the behavior of any creature you didn't build yourself).
Think of the car and driver. The driver controls a perception of the car's
position in its lane by the action of applying a turning-moment to the
steering wheel. Essentially all of the driver's actions are present only to
oppose the effects of independent disturbances on the car's path. Those
disturbances are unpredictable and invisible. Nevertheless, the driver's
actions correlate highly (negatively) with their net vector sum. But the
driver's actions correlate hardly at all with the visually-detected
position of the car on the road, which is the controlled perception.
How does this jibe with saying that action and perception are inseparable?
i'm not going to answer this question head-on at the moment. your
formulation of what's going on is in terms of separate action and
perception; it's difficult to answer that question within your
framework. let me try to get a step closer by formulating the
situation in my own terminology, which is interactionist but is
undoubtably somewhat idiosyncratic (there isn't ``standard''
interactionist terminology).
new england is covered in backroads--they are twisty, hilly, narrow,
and old, and are subject to nearly all possible weather conditions.
the way one drives them is to keep one's left wheels near, or quite
possibly over, the center line (if it is visible). essentially, one
drives down the middle of the road; this is becuase the danger of
running into someone else is much less than sliding ingloriously off
the road (when a highschool friend of mine ran off the road shortly
after getting his licence, all the adults clucked and opined that
obviously he had stayed too close to the edge of the road). the road
changes constantly with the weather; if there's been an inch of snow
already but nothing's been plowed, driving down the road means driving
in the tracks established by previous drivers, even if the tracks are
suboptimally placed, since following the tracks is safer than crossing
them.
in california, the roads are much newer, and since they'r never
plowed, they have little reflecting bumps on the painted road
markings. so driving down the road means being mindful of not
crossing the bumps (which are quite effective at rattling the car).
the roads in the santa cruz mountains are as twisty as and much
steeper than the backroads of new england (but they are better made
and have reflector bumps). there are no speed limits in the
mountains: if you know the road, you drive fast; if you don't, you
drive slow; if you're stupid and drive fast where you don't know the
road, you lose. what is crucial is that you stay on yr side of the
reflectors no matter what. if you cross the line, any sort of
encounter is a disaster. if you stay on your side, the only hazard is
rearending bicyclists.
i280 in san mateo county calls itself the world's most beautiful
freeway. i don't know about that, but it's certain one of the world's
best built, and the traffic's reasonably light. i have the sort of
commute that it's easy to zone out on, there's so little to pay
attention to. i do tend to notice roadsigns, though, and i know, for
instance, that after the woodside exit, the road's going to take a
gradual curve to the left, and i can account for that.
ok, that's my car and driver story. i mean it to illustrate that
there's no such thing as simply ``someone keeping a car on the road.''
there's always a whole lot more stuff involved. i have mentioned at
least the following factors: customs; geography; weather; road
construction; individual experience; various sources of information
about the road (eg, reflector bumps, road signs). i argue, first,
that to model this as a single process of controlling for the
perception of a single variable (or a bunch of them) would be
hopelessly baroque and probably impossible; second, if you did extract
from all this a common thing that we could agree was the essence of
``keeping to the road,'' then you're going to have to tell me where
you're going to put all the rest of the stuff that i've mentioned and
can produce pretty good evidence for.
the short interactionist story is that none of this stuff gets thrown
away, and it's not obviously locatable either Inside or Outside. the
long one will have to wait till my hands recover.