[From Bill Powers (951128.0805 MST)]
A rather neat example of spurious cause and effect showed up this
morning. The living room was pretty chilly when we got up, so we turned
on the solar heating register by turning up the thermostat to the
daytime setting. This was half an hour before dawn. Then a nice clear
day dawned, the sun came up, and Mary opened the curtains to let the
sunshine in through windows and the patio door (all Thermopane) and warm
up the room faster. Half an hour after that, the living room was cold
again. So we found empirically that letting the sun shine into the room
caused the room to get colder! Can you guess what mechanism was
responsible for this negative effect of heat input on temperature?
Bruce Abbott (9527.1650 EST)
Certainly at this stage of the game, reinforcement theory appears
to explain a wide range of phenomena of conditioning better than
PCT does.
Now, there's a challenging statement. Do you mean that reinforcment
theorists are of the opinion that reinforcement theory explains a wide
range of conditioning phenomena better than PCT does, or are you
asserting that it actually does? A great deal hinges on how you're using
the term "explain." Can you give an example?
···
-----------------------------------------------------------------------
Martin Taylor 951127 16:30 --
Did the ice cause the accident, did the drinking of the driver
cause it, did the failure to repair the headlight cause it, did the
fact that one driver delayed departure by 30 seconds to say
something to his wife cause it? I'd say that the question is
meaningless ...
Yes. These are the questions that used to occupy the time of the
Scholastics, and all natural philosophers before Galileo. My ancient
Brittanica says, at the beginning of the section on "Causality or
causation", "By the time [the word cause] came into English, in the 13th
century, the word already had a wide range of uses, some of them highly
sophisticated, since it was a key term in science, philosophy, and the
law."
The first section deals with "Sense I: Human agency." It says "Both
_causa_ and its Greek equivalents were used in legal contexts to refer
to the voluntary action of an agent for which he could be held
responsible." And then ...
Divorced from its association with the law courts, the word "cause"
is used for any action which an agent performs in order to bring
about an event or state of affairs (the effect), whether in nature
or in another agent.... In this sense, (Sense I) to cause an event
to occur is to perform an action with the expectation and intention
that the event will follow.
So in the sense of human agency, "cause" is used synonymously with the
PCT meaning of "control." The rest of the discussion is interesting from
that standpoint.
The second sense discussed is "Cause in nature." This begins to take us
back to the old philosophical diddles: "To discover the cause of an
event is to discover something among its temporal antecedents such that,
if it had not been present, the event would not have occurred." So if an
elbow knocks a vase off a table inadvertently, we can say that the cause
of the vase's fall was the movement of the elbow. Or should we say it
was the placement of the vase near the edge of the table? Or was it the
fact that someone asked what time it is, causing the person to draw back
the elbow in order to see the wristwatch? This was meat and drink for
the ancients, and evidently still is for their intellectual descendants.
The third sense of cause has to do with "Cause as explanation." After a
discussion of "weak links" being singled out as causes -- things which
could most easily be changed to change the outcome -- we get back to the
historical development from Aristotle to Copernicus. Aristotle gave us
the Four Causes: material, formal, efficient, and final. These causes
are answers to the question "Why is something what it is?"
The reviewer says, "The transition from explanations in terms of
efficient causes to explanations in terms of law was largely the work of
Galileo." And a little later, speaking of Newton, he says "But he always
regarded a force as an unknown somewhat (sic) which is the efficient
cause of observed motion."
The most interesting part of this discussion concerns the way science
eliminated Aristotle's fourth cause: Final Cause. The efficient cause of
a sculpture is the actions of the sculptor which shape the material. But
the final cause is "the purpose for which the thing is produced." In
short, science is here described in part as the process of finally
discarding purposive behavior as an explanation of anything. No wonder
it has taken close on 400 years for the behavioral sciences to recover
from that blunder! The elimination of purposive explanations from the
physical sciences was a great step forward, but the same step was an
enormous setback for the behavioral sciences.
-----------------------------------
Cat in box:
When it's in the box, the condition we all were talking about in
what Bill L. commented on, the question of how the cat got (back)
into the box really doesn't arise.
Perhaps you will remember my quotation from Thorndike, in which he
described the furious tearing at the walls of the box by the cat, the
extreme energy of the movements which continued for many minutes on end.
I think the question of the degree of error in the cat when first placed
in the box has a great bearing on the outcome. You seem to agree, later
in your post:
But since you brought it up, it's probably an important element in
why the cat wants to get out of the box. If there were no error
involved in its being in the box, as would probably be the case if
it wandered in of its own accord, there would be no reduction of
error when the door opened and nothing to learn. So, yes,
overwhelming force may be a component of the whole experiment, but
it was not what I was referring to in what Bill L was commenting
on, nor does it apply when the cat is actually in the box.
"Nor does it apply when actually in the box???" It is the enormous error
induced by forcibly placing the cat in the box against its will that
motivates all that follows, isn't it? Actually, in Thorndike's initial
experiment, the cat escaped by clawing at the door, eventually pulling
loose the loop of string holding the door closed. So it escaped in the
course of trying to get back to where it was before it was snatched into
this strange situation. We have to guard against tunnel vision when
looking at experimental results. What happens before and between
experiments may be as germane as what happens during them.
I didn't really suppose you to mean what you said, before, but it
could certainly have been misleading to quite a few readers, and it
definitely took _me_ quite by surprise, despite my years of
interaction with you. I'm not going to requote and comment,
because the above statement is succinct and all that is necessary.
The sentence you don't want to requote is
That is the main thesis of PCT: the ONLY reason for behavior is to
control perception.
From this sentence, you seem to have manufactured "All perceptions are
controlled by behavior." What took you by surprise was your own hasty
misreading of the sentence.
-----------------------------------------------------------------------
Rick Marken (951127.2000) --
I don't think EABers believe ANYTHING about the appearance of
control because they don't know what control (purpose) is.
While I'm definitely on your side here, you're fighting a losing battle
over the word "control." From what Bruce has said, EABers use this word
synonymously with "affect," "influence," or "determine," and have no
word at all for what a negative feedback "X" system does to its own
input variables.
Our problem is not that EABers don't know what "control" really means;
it means to them what it means. The problem is that they're missing a
_concept_, the one to which we refer in PCT by using the word "control."
They don't recognize the _phenomenon_ to which our usage of the term
refers. So all arguments in which the word control is used are at cross-
purposes: they argue as if it means "cause" or "affect" or "influence"
or "determine," whereas we argue as if it means selecting a state for
some perceptual variable and creating the actions needed to bring it to
and maintain it in that state despite disturbances tending to alter it
away from that state. EABers DO this kind of process all the time, of
course, in selecting how an animal is to behave, then producing whatever
actions are needed to make it behave that way. However, since they don't
have any word for this process, they perform it unconsciously, not even
realizing that they're doing something they can't explain. What they're
doing doesn't even exist in their perceptions at the verbal or
conceptual level. They can't talk about it and they can't think about
it.
It would really be nice to have a word of our very own which means only
and exactly "selecting a state for some perceptual variable and creating
the actions needed to bring it to and maintain it in that state despite
disturbances tending to alter it away from that state." Apparently,
people will simply not let us have the use of the word control for this
meaning, even though they have plenty of other words that mean exactly
what they mean when they say "control", and don't really need this extra
word. I think they're being very stubborn and selfish about this. What
they're doing is forcing us to invent a jargon term, so that when we use
it, they will not (we hope) automatically hear "influence", "determine",
and so forth and act as if that is what we said. Unfortunately, if we do
use a special term they won't understand what we're talking about,
because they lack the concept to which the term points. They will just
translate it right back into the terms for which they do have meanings:
affect, influence, determine, and so forth.
This problem is even worse than that. Suppose we show an EABer a demo in
which this X phenomenon is going on, and a block diagram showing exactly
how it works, and a prediction that predicts exactly what will happen
with a new disturbance. This should supply the missing concept, the
missing meaning, so that when we say "X" they will think of the
situation in the demo that they have just experienced. Unfortunately,
that is not what will happen. They will simply apply the explanation
that they assume applies to all behaviors, and see happening only what
they have already recognized as behavioral phenomena. They will see
discriminative stimuli and reinforcers and responses, and they will
weave these terms into sentences that have the grammatical form of an
explanation, with "because" inserted in all the right places.
And if we challenge them to make a prediction for a new experiment, they
will offer a prediction but not a quantitative one, not one that
actually predicts where the person will move the stick during the
experiment. When we complain about that, they will say that their
prediction fits the results perfectly, because they have no experience
with making "perfect" predictions of the the kind we get with PCT, and
have no idea of what we mean by "perfect." When we point to a tracing of
a model's prediction overlaid on actual behavior, they are unimpressed
because they simply don't see what they are looking at; they don't know
what it is about this tracing that we think is so important. They have
never seen results like that before. Sure, the subject followed the
instructions -- so what? (I have actually heard that comment more than
once).
I'm not hinting that Bruce Abbott has this problem; I don't THINK he
does. It's never easy to be sure, however; two people can be tossing the
same words around for hours and hours and never realize that the
concepts behind the words, in the two heads, are completely different.
It's much easier to detect disagreements than spurious agreements. The
main sign of a spurious agreement is that one person eventually says
something he fully expects the other to agree with, and meets with a
totally unexpected objection to it. The only way to handle this that I
know of is to hunker down and start going through the details one by
one: just what do you mean by "and?"
There may be no solution to this maddening problem. We can't make people
wipe their brains clean and start all over. Lots of people wouldn't want
to do that even if they could. Maybe we should start up that contest
again, although I've hated every suggestion for a new word, including
all of my own. Maybe we should try substituting the definition every
time we want to use the word control, putting in on a programmable key
or something. We can talk about a selecting a state for some perceptual
variable and creating the actions needed to bring it to and maintain it
in that state despite disturbances tending to alter it away from that
state system. After we have talked enough about theories that explain
the process of selecting a state for some perceptual variable and
creating the actions needed to bring it to and maintain it in that state
despite disturbances tending to alter it away from that state, and have
published papers distinguishing causal phenomena from selecting a state
for some perceptual variable and creating the actions needed to bring it
to and maintain it in that state despite disturbances tending to alter
it away from that state, and have said often enough that behavior is
selecting a state for some perceptual variable and creating the actions
needed to bring it to and maintain it in that state despite disturbances
tending to alter it away from that state, people will start begging us
to think up a word for that concept so they can understand what we're
trying to say about selecting a state for some perceptual variable and
creating the actions needed to bring it to and maintain it in that state
despite disturbances tending to alter it away from that state. Maybe
they'll even get so desperate that they'll let us use "control."
---------------------
P.S. I think you meant "that doesn't seem to faze them."
-----------------------------------------------------------------------
i.kurtzer (951127.2130) --
I know this might come off as glib but in all earnestness there is
no such thing as something being "unnatural" as when persons talk
about variables naturally changing as opposed to some other
"unnatural" manipulation. If you, as I do, assume that we are a
part of nature then what we do is as "natural" as anything else.
The only unnatural things are the things that do not, have not, or
cannot occur.
True. Reading the use in context, however, I would expect you to
understand "natural" independent variables as those that, when
manipulated, leave the system operating normally.
-----------------------------------------------------------------------
Best to all,
Bill P.