[From Bruce Abbott (960129.1545 EST)]
Bill Powers (960128.1800 MST) --
Bruce Abbott (960128.1400 EST)
The question being addressed is "what is a reasonable way for
organisms to be constructed if they are to survive?" ... If I say
that you must eat to survive, that is only a statement of fact. It
is not "a statement of faith in the purpose of evolution;" in fact
it says nothing about evolution at all. Go ahead, try it: don't
eat for, say, 40 days and see what happens!It is true to say that if an organism does not eat, it will not survive.
It is theological arm-waving to say that organisms have evolved to eat
in order to survive. It is factual to say that present-day organisms
have a certain organization. It is theological arm-waving to say that
evolution constructed them this way.
If an organism does not eat, it will not survive. Therefore, if an organism
survives, is must have eaten, and if it did not eat, it did not survive.
Nothing here but fact and logic. No arm-waiving or theology.
Darwinian evolution -- natural selection is the more proper term --
does not plan or construct. It only weeds out organismns that have
failed to learn to manage the variables on which their own survival
depends.
Natural selection is not "the more proper term" for Darwinian evolution,
Bill; this is like saying that negative feedback is the more proper term for
control. There are _three_ components to Darwinian evolution: (1) variation
(offspring are not always identical to their progenitors), (2) natural
selection (in a given environment, organisms possessing certain
characteristics are more likely to survive and reproduce), and (3)
like-begets-like (offspring tend to be similar to their parents).
The variations can be viewed as candidates and natural selection the
screening process. Nature proposes (variations) AND nature disposes
(natural selection), to borrow and adapt a phrase you also borrowed and
adapted. It's seeding and weeding, not just weeding.
And there isn't even any "it" here; the fact is that organisms
which do not control what happens to themselves with sufficient skill
fail to propagate, perhaps because they die or perhaps because they fail
to attract mates or for numerous other reasons. That is the only "force"
at work here, a simple matter of consequences. Nothing is doing any
selecting. There is success, and there is failure, and that is all.
There is nothing in the universe, not even an organism, that has a
reference signal saying "survive!"
Of course not.
Natural selection does nothing positive to change or improve species.
All such changes come about by spontaneous alterations in the genome --
cosmic ray hits, or conceivably internal random-change generators. Some
of the changes result in advantages to the species, others in
disadvantanges. The species with the advantages survive better than
those without them. And that is all that is left, when you speak of
evolution without any hint of purposiveness in it. If you will edit the
way you describe evolution so you don't use purposive terms, I will quit
bugging you about it. Either that, or explain how evolution could be
purposive -- that's an acceptable alternative.
I never meant to convey anything else but this blind uncontrolled process
you describe. If I state that it is through the process of evolution that
organisms have come to have the characteristics they exhibit today, would
you find that acceptable, or would you work to find some literal
interpretation of my words that would seem to convey purpose and intention.
You know, I can do the same to your words if I so choose. "The environment
proposes and the brain disposes?" Really? I can make you sound very silly
by taking that statement literally.
We know that in humans (and many other organisms), sugars generally
fall into the class of nutrients and alkaloids (when present in
sufficient concentration) generally fall into the class of toxins.
We know that newborn human infants will readily accept sweet
substances and will reject bitter ones. They do it the first time
these substances are encountered.
...
How is it that we humans come wired up to accept sweets and reject
bitter substances, right out of the box? One answer appeals to the
process of evolution as proposed by Charles Darwin.That is not the important question; the answer does not tell us
anything. To ask "why" we do this leads only the answer that our
ancestors did it and survived long enough to beget us and pass whatever
characteristics are inheritible on to us, and those human beings who did
not do it have probably died out, or at least are fewer in numbers than
those who did.
Asking "why" gets us thinking about what is required of the system if it is
to survive and thrive. It seems to me that identifying system requirements
is a big step toward system identification -- you know what you are looking
for. Sure, there are many ways this function could be carried out, but
you've narrowed the field considerably.
The important question is, "HOW are humans wired up, right out of the
box, so they accept sweets and reject bitter substances?" The answer I
would give is that they are equipped with inherited reference levels for
sensations aroused by these substances, and rudimentary control systems
that act (not very skillfully) either to increase or decrease these
sensations. The origins of this kind of organization are of only
academic interest; we could ask about origins no matter what kind of
system was inherited.
How they may be wired up should be pursued after one has some idea what the
wiring must accomplish. Then you have a much better idea what you are
looking for.
A much harder question to answer would be "How are organisms wired up so
that certain inputs can cause them to produce whatever actions will, in
the given environment at a given time, cause those inputs?" I know of no
organization other than a control system that can do this, whether it is
pre-wired or acquired.
And I am not proposing otherwise.
One thing we can be sure of: what matters is that the reaction to sweet
or sour substances must affect the experiencing of those substances in
the appropriate direction. Organisms that survive have to control the
kinds of substances that are admitted into their bodies. And this means
that they must contain, or learn, reference levels with respect to which
perceptions of the substances are evaluated. There are very few such
reference levels that are inherited at the level of motor behavior, and
very few complete control systems. Babies will quite happily eat their
own feces if they are allowed to do so. While they can detect intrinsic
errors, there is little they can do about them at first but create a
general energetic output having little organization.
Yes, and spit out things they don't like, or close their mouths and turn
their heads away when you try to insert more.
A consequence of this process of development is that current system
characteristics can be described in terms of function; one can ask,
"what does this characteristic, or this system, _do_ that enhances
the organism's reproductive fittness?"I think that this is a misguided question. It invites a purposive answer
where purpose is an inappropriate concept. Organisms do not develop new
capabilities in order to survive better. Better fitness is simply a
consequence of developing improved capabilities. The characteristic is
not developed in order to enhance reproductive fitness. That is putting
the cart before the horse.
Darwin's achievement was to show how a purposeless process could produce
organisms whose charateristics so nicely dovetail with their needs, as if
those characteristics had been designed into the organism in order to
achieve some purpose. If you understand this, there is no harm to be done
by asking what the apparent purpose of that characteristic is, and much to
be gained. True, once you know the mechanics of a system, you can observe
what functions it serves; in that sense mechanism is superior to function.
But questions about mechanism and function are both important, and it is
often possible to determine function before the mechanism has been
identified. Consider the question, "why does the arctic hare have white fur
in the winter?" Here are two answers:
1. Because reduced daylight leads to reduced production of a certain hormone
required for production of the colorant in the hare's fur. (Mechanism)
2. Because being white in the winter makes it difficult for predators to
detect the hare against the ever-present snow-white background. (Function)
Knowing #1, I can tell you what will happen to the hare's fur color as
sunlight decreases with the approaching winter. Knowing #2, I can tell you
why. And knowing Darwin, I can argue that the "why" does not imply that the
development of this characteristic was somehow guided by a goal or purpose.
How about answering this question about function, if you don't mind my
putting the horse _in_ the cart (so we don't argue about his whereabouts!):
Why do sensory inputs from various sources often produce a secondary
perception that appears to represent an hedonic dimension (values along a
scale of pleasantness)?
Well, you have at least one thing right: there is a close parallel
between evolutionary theory and reinforcement theory.Reinforcement is said to increase the likelihood of the behavior that
produces it. Natural selection can't increase the likelihood of
anything; it can only decrease the likelihood of an existing
organization in future generations if the organism fails to reproduce.
For a parallel, you would have to say that reinforcement suppresses all
behaviors except the one that produces the reinforcement.
I said the parallel is close, not perfect. In both cases there is
variation, there is some selective criterion, and this criterion affects the
probability the certain characteristics will be observed in the population.
That is, there is greater reproduction of those characteristics that pass
the selective criterion than of those that don't.
And you have not said _what_ is learned in the case of the organism. In
the case of the individual organism, what is learned has to be judged
relative to an internal standard. A consequence of behavior means
nothing until you know whether the organism is seeking it or avoiding
it. The seeking or avoiding is not determined by the consequence, but by
the internal standard and an organization that acts to reduce the
difference between the consequence and the desired consequence.
If you argue that evolution has resulted in that internal standard and
the organization that can match inputs to it, fine -- but you will be
arguing the control-system case, not the reinforcement case.
If I am, fine. Perhaps at some level the two views begin to say the same
things, just using different words.
There are no internal standards in reinforcement theory, are there?
I don't know. The old hedonistic viewpoint would say, if it feels good,
keep it up; if it doesn't, stop. Perhaps "good" and "bad" are states of a
signal or signals in comparison to some internal standard(s). At some
level, behavior would be organized to control those signals.
So sensory inputs give rise not only to undifferentiated trains of
neural impulses that constitute the perception of the controlled
variable, but to evaluative signals consciously perceived as
pleasure or pain? Now where have I heard that before . . . .I don't know. An "evaluative" signal could mean almost anything -- an
error signal, a signal that says "that was nice," a signal that says
"The author achieved a considerable degree of suspense but without
making us care about the outcome" or just about anything you want it to
mean. When you use general words like "evaluative" you're leaving so
many loopholes that you aren't really saying anything. In PCT, an
evaluation of error has to come from a comparison: the actual state of
an input compared with a reference state that defines what is the right
state of the input as far as the organism is concerned. Signals don't
just "have value;" they have values assigned to them. Any model needs a
mechanism for assigning value.
This comes back to the question I posed earlier: why to so many perceptions
seem to give rise to a secondary perception along an hedonic dimension?
This certainly isn't required for any individual ECU. Think about it before
you decide that I'm not really saying anything.
Regards,
Bruce