[From Bill Powers (951204.0640 MST)]
Bruce Abbott (951203.2005 EST) --
In EAB, to say that some environmental stimulus "determines where
the bird will peck" is only to say that there is an empirical
relationship between where the stimulus is and where the peck will
be directed.
I use the word "determine" to mean that given the antecedent(s), the
consequent can be none other than it is. In algebra, for example, we say
that a system of 3 equations in 4 unknowns is "underdetermined," meaning
that more than one solution is possible. A determinate linear system
consists of n independent equations in n unknowns, and there is at most
one solution. So I interpret the statement that an environmental
stimulus "determines" where the bird will peck to mean that given the
stimulus, pecking will always and necessarily occur in one and only one
place. This is the common meaning in science, as far as I know. If you
don't mean that, there are other words that should be used instead. For
example, if there are other variables that also contribute to the
consequence, then any one of them, including the manipulated one, can be
said to "affect" or "influence" the outcome. But no one of them
"determines" the outcome.
The stimulus is not assumed to "direct the peck toward itself"
(there you go with animism again) but rather, to innately "release"
a peck which the bird's brain mechanisms direct toward the
releasor.
The term "release" means to me a scenario in which a generator of action
is working, but some other force or mechanism is preventing the action
from taking place. When the restraining force or mechanism is removed,
the action then proceeds. This meaning is similar to "trigger," which
again implies a force that is being applied against a restraining
mechanism but is being held back. "Triggering," however, also means a
small influence that brings some process over the threshold of
spontaneous self-generation, as a spark triggers an explosion or as
releasing the gun's hammer results in the firing of the gun.
The "S-R" mechanism assumed to be at work here would be the
ethologist's "innate releasing mechanism (IRM), which acts as the
trigger; what follows CAN be open loop OR an organized, feedback-
directed pattern of behavior (the "fixed action pattern").
As you know, a control process is not "feedback-directed," nor does it
produce a "fixed action pattern." Control processes involve the
direction of perceived consequences toward a preselected reference
state, and they produce variable actions, not fixed actions. The action
seems fixed only when no disturbances of the perceived consequence are
permitted. The S-R mechanism that you describe tacitly assumes a
different model, one in which causality proceeds in a straight line, or
at least a strict progression of events from start to finish.
Since the bird can direct its pecks at things, it would be assumed
that the bird's brain is so organized as to perform the necessary
computations and corrections. How it does this would be left
unspecified, although as I suggested, I don't think anyone in EAB
would have a problem with the notion that a negative feedback error
correction system is as work. However, this would be _described_
in terms of external variables, as though a direct causal link
existed among the relevant variables.
As I tried to point out, accepting that the bird's brain is organized to
carry out the necessary computations and corrections is INCOMPATIBLE
with a description in terms of simple direct causal links. The only
conclusion I could draw would be that the person who goes ahead and
describes the situation in terms of external variables and direct causal
links has heard the words but has not grasped their meanings.
The only way you can go on speaking as if there
is an S-R connection is to have missed the point of agreeing thatthere is an error-correcting mechanism involved.
Well, you are the only one talking about an "S-R connection" as a
simple open loop system here, and that is where the problem lies.
Simply _saying_ that a closed-loop process is involved means nothing
unless you show _how_ it is involved, and reason while taking into
account the special properties introduced by the closed loop. If you
say that a stimulus triggers or releases a closed-loop behavior to
produce a fixed action pattern, you are contradicting yourself because
closed loop systems produce fixed action patterns only under very
restricted conditions which you have not mentioned: i.e., absence of
disturbances of the controlled consequence.
Again, you are employing an overly simple conception of what is
said to be going on. Certain properties of things, when perceived
by a bird in the right state (e.g., food-deprived in this case),
serve to release a pattern of directed pecking.
The alternative to this interpretation is that animals innately control
certain basic variables. A so-called unconditional stimulus is actually
just one of many possible events that can disturb one of these
controlled variables, and the so-called unconditional response is the
action by which the animal normally opposes those disturbances and
maintains the variable in its reference state. A conditional stimulus is
one that reliably precedes a disturbance, and it comes to be perceived
as a disturbance in itself.
If the action taken is itself an organized control process, then the
system in question is a higher-order system which varies the reference
signals for the lower (pecking) system.
This alternative explanation, its correctness aside, shows that by
contrast, the behaviorist explanation assumes a different model, one in
which incoming stimuli have special effects on behavior in a straight-
line manner, which is what I mean by S-R. The stimulus occurs, it
releases a behavior, and a fixed action-pattern is observed. The term
"release" has definite denotations that I have mentioned above, and
implies a kind of S-R model in which cause proceeds in a uniform way to
effect, as pulling the trigger of a gun causes the bullet to emerge.
For classical conditioning to occur, a stimulus must serve as a
signal for some event, and that event must serve to release a
pattern of behavior in reflex-like fashion.
That is the model assumed in the behaviorist account. In the PCT model,
there are no reflex-like processes, unless you take Dewey's view that a
reflex is actually a fast-acting closed loop (which behaviorists, as far
as I know, do not). So the behaviorist account is not strictly
empirical; it assumes a model of invisible processes that is indicated
by words like "releasing" and "triggering."
[A while back I mentioned that behaviorist descriptions focus on changes
rather than actual values, and you denied this. However, when you speak
of releasing or triggering "events," you are speaking of a pattern of
changes with a beginning, a middle, and an end. Other words for change
are onset, occurrance, and rate. When interval schedules are said to
open the loop, they do so only with respect to changes, not actual
values.]
···
-------------------------------------
Now back to the classical conditioning interpretations:
The US is presentation of grain in the hopper, the UR is pecking
directed toward grain in the hopper, the CS is the illuminated key
(which reliably preceded the presentation of the grain) and the CR
is pecking directed toward the illuminated key.
Why is the CR not pecking at the grain in the hopper? A diagram I have
seen looks something like this:
CS --------->-------
\
US -------------UR (CR)
In other words, the behavior produced by the US is initially an
unconditional response. When the CR is introduced, the SAME BEHAVIOR
becomes a CR because it is produced by the CS without the US being
needed. The dog initially salivates as a response to the food powder.
After conditioning, it salivates (the same behavior) as a response to
the bell without the food powder.
As you are presenting the situation, the initial response is pecking at
the grain in the hopper; this is the UR. The US, I presume, is the food
in the hopper. After conditioning, the same behavior ought to be
produced by the illumination of the key -- but you are saying that after
conditioning, a _different_ behavior is produced, pecking in a new
location (there is no change in the down-up pecking action itself). What
shifts the pecking location from the grain in the hopper to the key
where there is no grain? The shift in the location of pecking can't be
seen as classically conditioned, because the initial location is the UR.
The CS ought to produce the same location of pecking that the US
produced, shouldn't it?.
-----------------------------------------
The thrust of this post is really that the words used in behavioristic
explanations are not defined in an unequivocal way; they are used
connotatively, not denotatively. I recognize that in the initial stages
of any science, mathematical treatments may not be achievable. But
lacking mathematical treatments, we have to use ordinary language in a
more rigorous way than in informal speech. We have to settle on clear
definitions that have one and only one meaning: denotations. And these
definitions have to be laid out before any explanation using them is
given.
Do you think you could define classical conditioning in a denotative
way, so that given the list of definitions I would understand what is
conditioned in any situation exactly the same way you do?
-----------------------------------------------------------------------
Boy, you really ARE confused. Who said KEYpecking was innate? I
hope things make better sense to you now that I've tried to clarify
it!
Well, what you said was (951201.2105 EST)
But if keypecking is an innate response sequence released by
certain perceptions (under the right conditions), then the
procedure meets the specification for a classical conditioning
procedure.
Anyway, if keypecking is an innate response sequence released by certain
perceptions, it is a UR, not a CR. Isn't it? As I understand it, a UR is
an innate response to a specific stimulus. If it's a conditioned
response, it's not innate. Unless you are allowed to change your
definitions to suit each new situation.
If pecking is released by certain
perceptions under the right conditions, then sure, what happens would
happen. Obviously the perceptions are the right ones and the
conditions are right; you just said they are. But aren't those
assumptions being made out of thin air specifically to explain what
happens?
It's purely descriptive. One observes that, as you say, pecking is
released by certain perceptions under the right conditions, and is
directed toward the releasing stimulus.
I beg to disagree. One observes that pecking _happens_, and that it
happens in the presence of certain environmental conditions, but one
does NOT observe that it is _released_, nor that what releases it is
certain _perceptions_. Furthermore, one observes that the action carries
the beak toward an object, but not that it is _directed_ toward that
object, nor that it is a response to anything; neither is there any way
to observe what directs it. You are mixing theory with observation. This
is highly _impure_ description.
Why can't behaviorists get along without this obbligato of special terms
that bring all sorts of unnecessary connotations into the description?
What would be wrong with a bare description of what was actually
observed? It seems to me that behaviorists are always berating others
for a lack of objectivity, but when they describe objective happenings,
they can't resist peppering the conversation with terms designed to
further their own background views about what causes behavior (the
environment). Doesn't anyone teach behaviorists how to generate a
neutral description of observations?
In the past, one has observed that when a neutral stimulus is
paired in a certain way with one capable of releasing a certain
pattern of output,
Now you tell me: what is there in this sentence that is observed, and
what is imagined? Can you say the same thing without putting any spin on
it?
-------------------
One CAN think of any disturbance as A (not THE) cause of a control
system's action, yes?
If you want to think in terms of causes, sure. You can think of things
any way you want, as long as you haven't constrained yourself with
rigorous definitions and a preference for consistency. The relationship
of a disturbance to a control system's action is perfectly clear and can
be described mathematically. So in informal discourse, you can be as
sloppy as you please, I suppose, as long as you remember, and your
listener knows, what the actual relationship is.
Why can't biological systems be so constructed that certain
perceptions automatically set into motion a series of actions?
I suppose they could, in imagination, but if they were they wouldn't
work very well. The environment is not so free of disturbances that a
specific series of actions would always have the same consequence, or
anywhere near the same. Control systems have the nice property that they
can repeatedly create and continuously maintain a specific consequence
even though to do so they must vary their actions in unpredictable ways.
This is exactly what a real organism in a real world needs to be able to
do. If the sight of a bug in the air always triggered a left-turn
response in a bird, how many bugs do you think a bird would get to eat?
And if the sight of a bug always triggered an innate bug-chasing
program, how long would it take mama bird to bring her sticks back to
the nest she's building? Or do the bugs refrain from triggering the
fixed program when they realize that mama bird is busy doing something
else? You can't have it both ways. Either mama bird is initiating her
own programs in pursuit of whatever the current goal is, or the
environment is considerately turning its effects on and off so as to
cause all the behaviors that the bird needs to carry out to survive, and
doing so in just the right sequence.
The problem is that if you just observe behavior naively, you will miss
all the continual fine adjustments that the organism is making,
adjustments that are absolutely essential if the actions are to have
even approximately the same outcome twice in a row. It's all too easy to
confuse the OUTCOME of actions with the actions themselves, so you don't
realize how different the actions are each time you see the same outcome
occurring. The cause-effect concept of behavior simply overlooks the
active part played by the organism in seeing to it that the same effect
does indeed occur again and again. Without the organism's active
participation, without its continually striving to create specific pre-
selected effects on its own perceptions, the so-called "causes" would
have almost random effects, or no effects at all.
You keep speaking of reproducing ACTIONS. But control systems don't
reproduce actions; they reproduce consequences of actions, and to
reproduce the same consequence twice in a row generally requires
producing a DIFFERENT action each time.
As seen from the outside, it would appear that these perceptual
inputs were "causing" or "triggering" or "releasing" or "eliciting"
a response or even a pattern of activity, defined not in terms of
muscle contractions but in terms of the effect on environmental
variables (e.g., the key gets struck).
But each time the key gets struck, the organism is in a slightly (or
greatly) different position and moving at a slightly (or greatly)
different velocity; when the discriminative stimulus occurs, the
organism is at a different distance from the time before, and observing
some different aspects of its environment. How do the triggering or
releasing or eliciting stimuli know that their effects on the organism
must be different in each different circumstance? In order for the key
to be struck, the organism must make EXACTLY THE RIGHT changes in its
muscle actions. It's not enough simply to produce actions in the same
class as other actions that have had that same result in the past (as
Skinner proposed). Given where the organism is and what it is doing,
given what the environment is doing, repeating the actions that were
produced the last time will almost certainly result in missing the key.
But if you define the behavior as "keypressing" you will completely miss
the differences in the actions; it will seem to you that the organism
simply emits a keypress, nothing to it. Keypressing is not, however, an
action; it is an outcome of variable actions. There is no way in which
stimuli can produce keypressing again and again. The same stimulus would
have to have a different effect on the nervous system each time, an
effect exactly appropriate to all the disturbing influences that are
acting at the same time, so as to produce just the right DIFFERENT
muscle tensions each time. This fundamental problem with the idea of
environmental control was plastered over long ago, not to solve it but
to bypass it. But it is still here.
-------------------------------
By the way, I am getting the strong impression that you (and Rick)
believe that these reinforcement-theory explanations are the ones I
subscribe to.
It does sometimes sound that way. But even if you are simply describing
how others think, you seem to be defending their observational skills
and their verbal logic, neither of which I have found particularly
admirable. I see in behaviorism a STRONG bias toward interpreting all
observations, and describing them, as if the environment is the only
cause of behavior. All the terms are chosen with that bias: eliciting,
triggering, releasing, causing, stimulating, and so forth. All terms
that could give the organism an active role in events are dismissed or
mocked; the organism does not act; it responds. As I said before,
behaviorists seem unable to give a simple straighforward account of the
physical conditions and processes that are observed, without inserting
their own biases into every sentence. And when you report on the
behaviorist point of view, you don't raise any such criticisms, but seem
to use the same methods yourself.
Perhaps you conceal yourself too well. It would be reassuring to hear
you bring up some of these objections yourself, objections that would be
valid even if we had no PCT as an alternative view.
-----------------------------------------------------------------------
Best,
Bill P.