[From Bruce Abbott (2004.09.14.1435EST)]
Bill Powers (2004.09.12.0844 MDT)
Bruce Abbott (2004.09.11.1355 EST)–
I realize that ethologists were
much less doctrinaire about stimuli causing responses than other
psychologists, but their way of explaining behavior was no less confused.
How could it have been otherwise? If you don’t understand how control
works, organisms are very confusing. You think you’ve found a regularity
in the organism’s motor acts, and then, looking a little closer, discover
that a trivial-seeming change in the environment – the egg rolls a
little left instead of a little right – results in a complete reversal
of the motor act. How can the same situation result in completely
opposite responses? Or, if the little wobbles of the egg are the stimuli
and the adjustments of the beak are blind responses, why is it that the
responses so infallibly keep the egg on a path toward the goose? Why not
toward the North Star?
I appreciate your point that what may appear to be a fixed set of actions
may be “fixed” only insofar the “same” actions
(standing up, stretching neck toward egg, placing bill against back-side
of egg, etc.) can be carried out given the initial positions of the goose
and egg and the nature of the terrain between them. I must confess that I
don’t know what a goose would do if the egg were well beyond reach.
Would it get behind the egg and “shovel” it along in front of
it? I’ll see if I can find someone who knows.
Current ethologists acknowledge the fact that the behavior observed does
not always occur in the way most typically observed, and many of them now
use the term “modal” action pattern to suggest that the
described pattern is only the one most frequently observed.
Nevertheless, the goose’s egg-retrieval behavior can be accounted for in
principle by assuming that the goose has a higher-level control system
for which certain visual inputs act as a disturbance. (It is tempting to
suggest that the goose is controlling for “egg in nest” but
this describes only the usual consequence of the behavior as we humans
perceive it.) Error in this system results in the setting of references
for other perceptions at lower levels and the end result is typically the
sequence described, ending with the goose back on the nest with the egg
hidden beneath the goose.
Ethologists have been clever in designing experimental tests to identify
what specific features of the input set the pattern in motion, and in
some cases to investigate the development and nature of control. One of
the earliest experiments of this latter type was conducted by Douglas
Spalding in the late 19th century, who had noticed that the pecking
accuracy of chicks improved rapidly in the first couple of days after
they emerged from the egg. Was the improvement due to practice or simply
to maturation? To answer this question, some chicks were prevented from
practicing eye-beak coordination by placing hoods over their heads after
helping them out of the shell, whereas others were allowed normal visual
experience. When the hoods were removed after three or four days, the
formerly hooded chicks pecked as accurately as the ones that had never
been hooded. Eckhard Hess more recently improved on this experiment by
placing goggles on the hatchlings instead of hoods. The goggles displaced
the images of things seen by the chicks seven degrees to the right.
The chick’s pecks at a target “seed” were initially scattered
and displaced from the target. After four days the pecks were
tightly clustered and still displaced from the target, by the same
average amount. The chicks were unable to learn to adjust their aim. This
suggested to Hess that the coordination of eye and beak required to hit a
target is innate.
It
may be, as Rick has guessed, that snatching the egg away failed to stop
the behavior simply because the sudden loss of resistance to motion
arising from the sudden removal of the egg simply constituted a
disturbance that the goose’s bill-positioning system could not compensate
for quickly enough, but it is interesting that in most cases the goose
will also complete the sequence by sitting down on the non-existent egg
(or so I have been led to believe).The best bet in reading descriptions in behavioral literature is this: If
you haven’t seen it for yourself, don’t believe it. One “could
say” that in the movie clip, the bird “carried the movement to
completion”. Aside from the fact that this is a clear instance of
saying that the movement is what was triggered, not the
consequence of getting the egg back in the nest, I saw the same movie and
I would not say that the movement was carried to completion. What is
completion? The head located downward and to the side, or stuffed under
the breast in front? In the three instances shown, one mildly different
and two very different endings were shown. If one of them is seen as
“completion” of the act, the others can’t also be completions
of the same act. If it’s not the same act in all three cases, it’s not a
“fixed action pattern.” And, Mary points out, should we assume
that the bird’s head remained in the final position for the rest of its
life? What is “completion”? Isn’t it the point where the
observer stops paying attention and starts looking for a new act (or
stops the movie)?
I went back and reviewed the video clip and it is apparent that we don’t
always see the entire retrieval sequence due to bad editing. As to what
constitutes completion, look for the goose sitting on its nest with the
retrieved egg beneath the goose. The act is complete when the error has
been removed. (This reminds me a bit of a novelty toy I once saw, which
consisted of a box with toggle-switch on its side and a warning written
beneath the switch not to flip it. So, of course, that’s what you do. The
lid of the box then opens and a mechanical hand emerges. The hand reaches
down and flips the toggle switch back to the “off” position,
and then disappears back into the box as the lid closes. The sequence
ends with everything back the way it was. The mechanism was a simple
control system, although unlike the goose’s mechanism, the individual
elements of the sequence were not themselves accomplished by
control-systems.)
Explanations in the absence of
knowledge are not very helpful; I’d rather hear how we might go about
determining what the goose is actually controlling. The first thing we
have to realize is that the goose does not live in our perceptual
universe. If there were a person sitting there we could at least
extrapolate from personal experience to begin guessing what perception
the organism is trying to control. But this is a goose, and how an oval
(i.e., egg-shaped as we see it) object looks to a goose is far from
obvious, if it is even conceivable to a human being.
I addressed this issue in my post.
The
point I am making is that the ethologist’s conception of the “fixed
action pattern” is behavior that has a recognizable pattern to it,
even though it may begin from different positions and therefore require
different muscle outputs.You are saying what I said: that ethologists (and many others, I add)
confuse motor actions with their consequences. The consequences, the
patterns of which you speak, are not consequences of the motor acts, but
of the sum of motor forces and all other independent influences that
physically affect the same consequence. If you remove the independent
influences, the same motor act will have different consequences.
Therefore there is no possibility that there is a stimulus that affects
only the organism that can repeatedly produce the same consequence when
the independent influences change.
I don’t see how you conclude from this that ethologists confuse motor
actions with their consequences. They have been very careful to describe
what is repeatable in a given pattern of activity without making the
claim that this is accomplished via the same detailed set of muscle
contractions each time. This is the way most of us describe the behavior
we see: “John reached for the glass of water,” meaning that
John was observed to move his hand toward the water while reorienting the
hand and fingers as required to grip the glass. This description doesn’t
tell us how John managed to accomplish this pattern of movement, and by
the same token it does not offer a theory movement couched in terms of a
fixed sequence of muscle contractions.
Of
course, there’s a far simpler way to account for this flexibility, and
that is to assume that each behavioral act is the product of a control
system organized to accomplish certain goals.Of course I agree. I guess I get impatient with what sounds like making
excuses for older theories, instead of just concluding that they were
wrong and dropping them.
I’m not, as you say, “making excuses for older theories.” My
reason for bringing up the so-called fixed action patterns again was to
offer empirical support for the proposition that there exist in an animal
such as the goose control systems that are innately organized and which
carry out functions that are useful in the life of the animal.
Bill,
you really need to read the actual account written by these guys before
you start making assertions about what they did or did not notice,
realize, or understand about the behaviors they observed.I have read them, though I admit that I’m speaking about my general
impressions rather than exact recollections.When we went over the goose’s
egg-retrieval program (as I would term it), I noted at the time that
Tinbergen and Lorenz described how the goose moved its bill from side to
side as necessary to keep the egg balanced on the bottom-side of its bill
as it rolled the egg toward its nest. These investigators may have seen
the goose’s behavior as machine-like, but only in the sense that the same
sequence of acts could be observed on each occasion, and in each nesting
goose, and observed under the same general conditions (egg visible
outside nest).But that is exactly my point. The sequences of acts were different on the
three occasions in the film clip, and I would wager they were different
on every occasion and with every different bird, and that not even the
situation of “egg outside the nest” was the same on any two
occasions. “Oh,” you might say, “You’re being picky about
the details – it was the same general behavior that occurred in
the same general situation.” Yes, I would reply, when you
observe through a perceptual filter that removes the differences, you see
no differences. But I say there were differences, that they were not
trivial, and that the exact way the behavior changed with each variation
was exactly consistent with the hypothesis that the bird was trying
to stuff the egg under itself. There was no fixed action sequence, except
in the perceptions of observers who believed they were seeing fixed
action sequences. I did not see any. What I saw were actions that were
highly variable, occurring in three different sequences, and that (save
for the overwhelming disturbance applied by the experimenter) were quite
successful in the consistent end they achieved by different
means.
Picky? Not at all; skepticism is good, as it leads to critical
examination of the evidence, and it’s one of the things I appreciate
about your thought process. However, you do have me wondering here
whether you might not be going too far. Seeing
“repeatable” patterns in things always requires ignoring a
certain amount of variability. You could argue that there are no such
things as human beings as a general category; after all, no two are
identical. So here’s a hypothetical: If you saw people introducing
themselves to each other and shaking hands, would you claim that there is
no such thing as a handshake, because everyone did it a little
differently? I suspect you would not. Although motions may differ to some
degree, they tend to occur on similar occasions, serve the same social
function, and everybody involved, including most onlookers, recognizes
them as handshakes. It’s the same with courtship rituals in
grouse.
They
went on to test the visual characteristics of the egg to determine which
of them were necessary and sufficient to trigger the retrieval sequence,
with some surprising results (e.g., an empty beer can worked almost as
well as an egg!).As I say, our perceptual world is not the goose’s perceptual world. Why
should we be surprised that to a goose, there is something about an empty
beer can (or in other experiments, a smooth stone or a darning egg) that
is just like a goose-egg ? To some doting human beings, a Chihuahua
apparently looks and acts just like a human baby. We do not know what
perception the goose is controlling. Even to say that it is trying to get
an egg into the nest is to substitute our vastly more complex perceptual
systems for the goose’s. Maybe what it controls for is feeling a certain
size of smooth lump under it.
Yes, and that’s exactly the viewpoint ethologists adopted. Consequently,
they have invested considerable efforts in teasing out the relevant
perceptions.
(By the way, I would much rather
hear about reports from people who had not been drinking beer before they
did the experiment)
You would be amazed by what else the goose will treat as if it were a
stray egg, and retrieve it into the nest. And by the way, once the goose
sits on the beer can, something apparently doesn’t feel right, as the
goose then stands up again, lowers its head to get a look at the object,
and then uses its bill to kick it out of the nest. Apparently the
criteria for what constitutes an “egg” do not include the
tactile sensations produced by a beer can’s sharp corners.
And, of course “the”
retrieval sequence you mention is an artifact of human perception.
Actually there were different sequences of action every
time.
I would not go so far as to say that there is no repeatable retrieval
sequence to be observed. In class I show a video of a herring gull
(which like the goose nests on the ground) land near its
“scrape,” walk right past its own egg (which has been moved a
couple of inches outside the scrape), snuggle down in the scrape as it
always does (as if it were seating its breast over the egg), then notice
that its egg is in fact lying on the ground outside and go through
essentially the same motions as the goose to bring the egg into the
scrape. The method used by the goose and gull to retrieve their eggs
depends on initial conditions; if the egg were further away, the birds
would have to walk to the egg in order to reach it, for example. If what
the gull or goose has is a control system whose functional result is
usually to retrieve loose eggs, it is a program sophisticated enough to
deal with these different situations as required. However, you won’t see
one bird pushing the egg along with the bottom of its bill, in the
direction of its body, another pushing with the top side of the bill and
away from its body, and yet another bumping the egg along with its chest.
Only the first method is used, and it is a method that the bird did not
learn to use.
Of
course, these ethologists weren’t thinking at the time about control
systems, but it is quite evident that these are patterns of activity
emerging as the animal controls certain of its perceptions at perhaps the
sequence or program level. It is just as clear that these systems are
inborn, as they emerge even when the animal has been deprived of the
opportunity to learn them, and are seen in the same form among all
members of the species.What is inborn? You are going on as if it had been established that there
really are fixed patterns of activity, when the observations show that
only the goal is repeatedly achieved: getting the egg under the
body, to put it in human terms. The highly variable sequences of action
are at every instant whatever is required by the shape of the egg, the
current terrain, and the effects of previous actions to keep the egg
moving closer to the body. You say it is clear that “these
systems” are inborn, as if you have established that some fixed
action pattern have been proven to exist. But you haven’t
established that, so there’s no point in going on as if you
have.
You are placing too much emphasis on that word “fixed.” Some
examples of FAPs are more stereotyped than the egg-retrieval program is,
but even in the latter case, it is possible to say with confidence that
when incubating eggs, these birds do have a mechanism that functions to
retrieve loose eggs, that the methods used to accomplish this are
available to the bird on the first occasion to use them, and that the
retrieval act is set in motion when the bird’s visual perceptions include
certain visual features of nearby objects (resulting in a perception that
does not match its reference).
I don’t see that PCT would have any problem accounting for such behavior;
all that is needed is to construct a proper model based on appropriate
empirical data. And, contrary to something you stated in your previous
post, there is nothing here to require that the control systems involved
be unique to this particular program. Just as with any other
control system in the hierarchy, these special control systems would use
the same perceptual and motor systems that enter into the control of
other perceptions, setting references for the lower-level systems as the
means of accomplishing their own goals. The particular controlled
perceptions would be derived from the ordinary visual, tactile, and other
sensory systems connecting into an input function that would create the
controlled perception from these inputs, appropriately filtered.
So, in my view, the only real difference in our positions lies in my
assertion that these are inborn, special-purpose control systems, whereas
you want to deny their existence. I must confess that I am mystified as
to why you seem to be so resistant to accepting the idea that these
systems do exist. In my opinion ethologists have done a superb job
of demonstrating that they do and have even identified some of the
critical features of sensory input that define the CV in these control
systems. The question is not whether these patterns of behavior occur,
but how to best model the system that generates them. Developing such
models will require additional empirical work, of course. It seems to me
like a great way to demonstrate the value of PCT.
If
all you care about is to explain the behavior in terms of mechanism, then
appeal to the possible “survival value” of a given behavior or
set of them is unnecessary.Correct, and that is indeed all I care about. Of course I have proposed
ideas about the mechanisms by which survival value might actually come
into play, but they don’t rest on the assumption that just because some
behavior might promote survival, it magically occurs. It is just too easy
to rationalize any theory, even opposing theories, by appealing to
survival value. Why does a duck sit on an egg so heavily as to break it?
Oh, that gives the remaining eggs a better chance of being taken care of
and surviving. Why does a duck work so hard to keep its eggs from being
broken? Oh, that’s because the more eggs that remain unbroken, the more
likely the species is to survive. As Bruce G. says, we love to to make up
stories to explain things, to make them seem rational.
I indicated in the text that followed how such “stories” can be
tested empirically.
Given certain starting values and a set of time-varying inputs, the
machine does what it does because of its internal structure. But this
will not do (by itself anyway), if what you want to know is
“why” this mechanism exists rather than some other, or perhaps
none at all.You’re thinking of an old-fashioned industrial-revolution machine, not a
machine with sensors and computers and actuators and power supplies and
self-repair facilities and the ability to evolve toward better and better
control of what happens to it. The idea of machines as devices that
simply convert time-varying inputs into outputs is outdated by at least
half a century. When we think of “mechanisms” of behavior we
should not think of a ribbon of clanking punch-cards flowing through a
Jacquard loom in some dark satanic mill, but of a little stick man on a
computer screen reaching out to touch a target that you move
unpredictably in three dimensions, and doing it without any detailed
instructions at all.
As should be clear by now, I’m not thinking that way at all. Let me
rephrase that sentence: Given certain starting values (initial
conditions, parameter values), and a time-varying set of inputs that are
transformed into the controlled perception, the control system does what
it does because of its internal structure. Perhaps I didn’t say it
clearly enough, but that’s exactly what was in my mind when I wrote the
earlier version of that line. Go ahead, tell me that this is a
description of “an old-fashioned industrial-revolution
machine.” I DARE you. In fact I double-dog-dare you!
(;–>
“Why” such complex
machines exist in nature is exactly the story of “how” they
came into existence, and we can begin to discern the how as we learn more
about how they work. Just saying “survival” tells us nothing:
all the living organisms we have studied have survived, so we can say of
any imagined process whatsoever that it must have promoted survival above
all else, even if we don’t know the theory is correct. But most important
of all, what does anyone know about the survival value of any particular
action or pattern of action? Nothing, unless we understand the underlying
mechanisms of behavior. Otherwise it’s all mere
guesswork.
I agree that it is not enough just to hypothesis that some bit of
behavior exists because it had “survival value” during the
evolutionary process, and I am certainly not asserting that. However, it
is possible to figure out what some particular mechanism “does”
by making an appropriate set of observations. Description of a radio:
this little button turns it on and off, this dial tunes in the station,
this one adjusts the volume. I don’t have to know anything about how
those dials and buttons affect the inner machinery of the radio; for all
I know it happens by magic. Does this thingie on the airplane have
anything to do with flight? Well, let’s try flying the airplane without
it. Oh, no, I can no longer control the pitch! What did you say that
thingie was? Oh, the elevator? Having an elevator on the aircraft
has great survival value for the pilot who attempts to fly the aircraft.
You say there is no way to tell? Balderdash! (I’ve been looking for an
excuse to use that word.)
Please recognize what I am NOT saying. I am NOT saying that knowledge of
mechanism is unimportant. Once you understand the mechanism, you can
predict how it will behave under any given set of conditions (so long as
these don’t break the mechanism), and that is a powerful ability. But
absent such knowledge, you can make a good start by observing how the
mechanism behaves under given conditions. Any proposed mechanism will
have to generate those behaviors under those conditions before it can
even be considered as a potentially correct model.
Darwin’s
theory of evolution by natural selection offers a way to answer that
latter question: can one show that the mechanism has survival value? Or
to put it another way, what necessary function (or functions) does it
serve in the life of the organism that contribute to the organism’s
success in reproducing more of its kind?I think this proposition is backward. You might explain survival by
showing how the operation of a known mechanism would promote survival,
but without knowing how the mechanism works, you can’t use survival as
evidence that your proposed mechanism is correct. All that natural
selection proves is that all the organisms now present have survived, and
those that are no longer here have not survived. It has not even been
shown that maximizing reproduction promotes the survival of a species;
one can suspect exactly the opposite in certain cases, such as our
own.
What has “survival value” depends on the environment in which a
given property of the organism exists, including other parts of the same
organism. Maximizing reproduction does not necessarily promote the
survival of a species because the species is not the unit on which
selection acts. What may be good for the survival of an individual’s
genes may not even be the long-term survival of the individual, and even
what is good for the survival of the genes at one point may suddenly
become bad at another, and the change happen so fast and extinction
occurs before the process produce the necessary changes.
Everyone
can play the game of “just so” stories to “explain”
why some mechanism has survival value, and lately it has become the vogue
to criticize functional explanations on this basis.Isn’t that a partisan way of saying “More and more people are
starting to notice the flaws in the functional approach to
explanation?”
No. Since when do you, of all people, count the number of people in favor
of an idea as an index of its merit?
What
the critics have ignored is that it is often possible to test these
proposed explanations empirically. For example, it can be shown that in
birds that perform elaborate courtship rituals, getting it right makes
the difference (for the male at least) between fathering many offspring
and fathering few or none at all.But that is still a just-so story. I would conclude that females who have
become hypercritical about the courtship of males are lowering fitness of
the species by making reproduction less frequent than it would otherwise
be. Clearly that is the immediate effect of not accepting the first
applicant, isn’t it? Once you see such alternatives, you then have to
extend the rationalization, saying things that are even more fanciful,
like “Oh, but the males who put on the best displays show that they
have better memories or more energy or … blah blah blah.” You
can’t lose, you can only get tired and drop out of the
argument.
It is not still a just-so story; you have shown that, as currently
organized, the bird’s reproductive success depends on properly performing
the ritual. You are correct in saying that such displays may ultimately
lower the reproductive fitness of the species, but again, the species is
not the unit of selection. Evidence suggests that many animals have
entered an evolutionary dead end developing ever-more stimulating (to the
opposite sex) features that win the competition for mate over competitors
with smaller or less elaborate versions.
Your
strictly mechanistic approach, given enough data, could work out the
exact nature of the control systems whose actions bring about these
elaborate sequences of head bobbing and weaving, strutting, clucking, and
so on, but in the end you would still be left wondering why in the world
these birds are doing all that.My “strictly mechanistic” approach is simply to demand an
explanation for how things happen before I believe any theory about them
or even believe that the phenomenon has been described correctly, or that
the relevant aspects have even been noticed. I would guess that all the
bobbing and weaving has no positive function whatsoever; it’s simply a
side-effect of whatever the critical mechanism is. I assume that the
critical mechanism would work even better without all that wasted effort,
but as we know, natural selection always stops when it’s just barely good
enough.
Again, I agree that one should be demanding. Whether “all that
bobbing and weaving” might better be replaced by some mechanism that
would do the same job without all the waste of time and energy is beside
the point. The birds have to do it, because random variation and natural
selection has led their evolution down this path. Empirically, one can
show that performing this dance correctly is absolutely necessary if the
bird is to get the opportunity to copulate and therefore, for this kind
of bird, impacts on the frequency with which its genes will be
represented in the next generation.
Heck, maybe it’s just a fad, like
the Funky Chicken courtship ritual. The birds do it because it’s kind of
fun, and put off mating until they’re tired of the fun.
I like that – the Funky Chicken courtship ritual! And maybe – just
maybe – the geese retrieve their stray eggs because they enjoy playing
croquet and this is the closest they can come to it!
Come on, Bill, you know you really want to agree with me. Why not
just give in? You will feel so much better . . . (:–>
Best wishes,
Bruce