A simple form of learning

[From Bruce Abbott (970716.1015 EST)]

Bill Powers (970714.1953 MDT) --

Bruce Abbott (970714.2005 EST)

What additional information do you need?

We need to understand more of the whole system. As you said, for example,
there are other neural pathways than the simple one you described, and they
could well be acting on the basic loop. In more complex organisms, it's
clear that there are connections from sensory nerves to muscles, some
through a single synapse and others through internuncials. But these loops
can't explain behavior all by themselves; we can't even say what they will
do without knowing how their side-effects are treated by higher systems
which contribute to the reference inputs (alpha and gamma) of the spinal
loops, and perhaps even alter the chemical environment of these loops.

We are probably still some way from being able to provide the complete
description of the system you demand. However, this need not prevent us
from examining and learning from the information that has been brought to
light thus far. We know from observation, for example, that touching the
siphon brings about contraction of the mantle shelf, and that this action
diminishes with repeated identical touches of the siphon, if they occur
frequently enough. We know that this diminishment of response amplitude is
mediated by changes taking place in the axon terminals of the sensory
neurons, at their junctions with the motor neurons that stimulate mantle
contraction _and_ with interneurons which themselves terminate on those same
motor neurons. We know that the response amplitude diminishes because less
neurotransmitter is released by the sensory neurons with each impulse, and
that this effect is mediated by structural changes, including inactivation
of calcium channels in the axon terminals and (in the case of long-term
habituation) a reduction in the number of synaptic junctions. This
experience-dependent plasticity of neural conduction may be a prototypical
example of the cellular changes that take place during learning.

There are other connections (of course), but these are not relevant to
the phenomenon, which has been traced to changes taking place in the
sensory cell itself when it is repeatedly stimulated at relatively
short intervals.

But what is _causing_ those changes? The causes could well be outputs of
higher systems of either neural or chemical nature. There's no way to
understand a complex system until every variable has been traced to the
output of some system function, or has been determined to be independent
(originating from outside the system). The picture you present is full of
dangling variables.

The picture I presented was a highly simplified one, as it must be in this
forum. The changes I noted have been found to occur as described both in
the intact system and in "reduced" circuity in which the basic elements are
tested in vitro. Thus, while higher systems could be involved in the intact
organism, the mechanism described works as advertised without them.

I like that description, but it is not quite accurate: the loss of
amplitude response may persist for some time and attenuate the >amplitude

even to widely spaced stimulations that by themselves would >not have
produced measurable habituation.

You're right. The effect is more like that of an automatic gain control
with a fast "on" time and a long "off" time constant.

Yes, that's the way I've been thinking of it.

A feedback control mechanism with a fixed reference level of zero is
indistinguishable from a stimulus-response system.

But it can still be a significant part of a _system_ of control, yes?

No. Relative to an organized hierarchy, for example, it is like an
independent parasite; its behavior is entirely localized and not subject to
manipulation by the rest of the hierarchy. It's like a separate organism,
not part of a whole system.

Ah, but you are assuming a particular form of organization. There are other
behaviors in which Aplysia contracts its mantle musculature, and separate
neural mechanisms are involved. These mechanisms are interconnected so as
to maintain singleness of action where the different behaviors would
interfere. We have similar interconnections: e.g., you can't sneeze without
closing your eyes. The sneeze reflex has priority, so if you needed to see
at the moment the sneeze occurred (to catch a ball, for example), too bad.
In a simple organism like Aplysia, several different control systems may
operate at the same level, through some of the same mechanisms (e.g.,
mantle-contraction muscles), with mutually inhibitory interconnections
serving to organize output so as to minimize conflict.

There are ways to put functional hypotheses to test. If I deduced
that eyes are for seeing, I suppose you would argue that this is
poppycock. Perhaps they are there merely to provide support for the
eyelids.

Since they do both, you can't argue that they do NOT provide support for
the eyelids. Eyes provide visual signals, eyes provide support for eyelids,
eyes provide signals from an actor to an audience about emotions, eyes
provide one element of crying, and so on. But seeing did not exist, none of
these "functions" existed, before eyes existed; it was the other way around.

This "functional" stuff is an awkward way to talk about purposive or
goal-driven behavior when there is actually purpose, and it's a way of
asserting purposes or goals when there are none. What is the "function" of
a ruler? Is it to draw straight lines, to measure distances, or to prop a
window open? The function of a ruler is whatever function it is made to
serve by the system using it.

No, I think you are wrong, although I agree that it is easy to confuse
function and purpose given the way our language is structured. Objects can
serve various purposes, it is true (it has been said that outdated personal
computers make good boat anchors). However, the functional question asks
what is the role of some particular component within a functioning system.
The little knob on my car radio could serve as a nice hook from which to
hang a picture, but its function within the radio is to set the volume.
Within my lowest-level control system, the function of the muscle spindles
in my biceps muscles is to provide a perceptual signal (about muscle length
and length rate of change) to the motor neurons serving that muscle (and
others). We deduce these functions by examining how they affect other
variables in the system. In the same way, we can observe what effect
Aplysia's mantle-contraction reflex in response to touch has on important
variables that impact on Aplysia's ability to survive and reproduce. We
observe that fish and other aquatic organisms attempt to nibble on Aplysia's
gill. We observe that the resulting damage to the gill often leads to
Aplysia's death. We observe that rapid contraction of the gill into the
mantle shields the gill, preventing those organisms from gaining access to
it and eating it. We infer that mantle contraction to touch is a protective
reflex (or at least that protection of the gill is at least one function of
the reflex). An we can easily test this inference by disabling the reflex
and observing how this manipulation affects Aplysia's survival.

In this way, I can determine what the reflex does for Aplysia (function),
without ever asserting anything about teleological purpose.

The only way I would ever agree that Aplysia retracts its siphon to keep
its gill from being injured would be if you showed me the reference signal
for degree of injury, and a perceptual signal to go with it, and the
control system that closes the loop by operating the right muscles.
"Functional" comments about siphon retraction are an opinion of the
observer and have nothing to do with how the system works or how it got
that way. The "functional" approach is subjective.

See above.

. . . And by the way, since this is a perfectly
reproducible and predictable phenomenon, no reorganization is involved.

In your own example of reorganization, in which the gain of a control system
is varied so as to establish control, the outcome is also perfectly
reproducible and predictable. So by your logic, it isn't reorganization either.

I agree that habituation of Aplysia's mantle withdrawal response isn't
reorganization as you have defined it, but for a different reason.
Reorganization is supposed to involve persistent error in intrinsic
variables, and is mediated by a separate control system, apart from the
control hierarchy. Thus, as you define it, reorganization is not an outcome
(e.g., change in synaptic connections) but a particular mechanism through
which that outcome is achieved. Because the changes in synaptic
connectivitiy observed in Aplysia do not appear to depend on that mechanism,
we are not observing reorganization, _by definition_.

However, we _are_ observing the proximal _result_ that reorganization is
supposed to produce in the nervous system: change in synaptic connectivity.
If the reorganization process you envision exists in the human nervous
system, the synaptic changes it produces are likely to be mediated by the
same or similar biochemical mechanisms as observed in Aplysia's sensory
synaptic terminals.

At the very least, the habituation mechanism discovered in Aplysia shows
that learning (structural changes in the nervous system due to experience)
can occur in the absence of reorganization; thus reorganization is not the
only form of learning. Whether reorganization as you define it exists at
all remains to be demonstrated.

Regards,

Bruce

[From Bill Powers (970716.1212 MDT)]

Bruce Abbott (970716.1015 EST)--

We need to understand more of the whole system.

We are probably still some way from being able to provide the complete
description of the system you demand. However, this need not prevent >us

from examining and learning from the information that has been >brought to
light thus far. We know from observation, for example, >that touching the
siphon brings about contraction of the mantle shelf, >and that this action
diminishes with repeated identical touches of the >siphon, if they occur
frequently enough. We know that this >diminishment of response amplitude
is mediated by changes taking place >in the axon terminals of the sensory
neurons, at their junctions with >the motor neurons that stimulate mantle
contraction _and_ with >interneurons which themselves terminate on those same

motor neurons. We know that the response amplitude diminishes because
less neurotransmitter is released by the sensory neurons with each
impulse, and that this effect is mediated by structural changes,
including inactivation of calcium channels in the axon terminals and >(in

the case of long-term habituation) a reduction in the number of >synaptic
junctions.

OK, this is the phenomenon. But as I said there are loose variables
dangling -- everywhere you say "because," there is a cause described which
is not described as an effect of something else.

It could very well be that something like inactivation of calcium channels
is caused by some internal interaction inside the neuron itself, rather
than by some extraneous process. If so, this property will always be seen
no matter what is causing the sensors to respond, or in what context. For
example, the calcium channels could become deactivated simply as a
consequence of the signals that are generated by that same neuron, in which
case we would have a closed loop and that variable would no longer be
dangling.

It is vital in analyzing any system to account for all variables, even if
you have to guess. If you leave any variable unaccounted for, there is a
chance that it is affected by some other route, through some other system,
and the local effects you observe might be extremely misleading even if
they're observed accurately. Going _in vitro_ doesn't help; you can analyze
a muscle-response to incoming signals _in vitro_, and observe exactly the
same relationships you see _in vivo_, but this will never reveal the
control loop that the muscle is part of.

Maybe what I mean will be clearer with an example. Suppose that p =
f(x,y,z). I tell you the form of the function f, and the values of x and y,
and then ask you to compute the value of p. Can you do it? No, you can't,
because z could have any value at all. You can't even estimate p unless you
know z, too -- or at least guess at how z is going to vary. In analyzing a
system, you have the same problem multiplied. You may know some of the
relationships in the system very well, but if there are any dangling
variables, there is simply no way to predict what the system is going to
do. Any prediction would be empty of meaning. Or worse, the "prediction"
would come from having observed how the system really behaves.

This experience-dependent plasticity of neural conduction may be a
prototypical example of the cellular changes that take place during
learning.

Yes, it might, but we won't understand what's really happening until we can
say _what_ experiences produce the changes, and _how_. I don't fault the
naturalistic approach, but if we want models we have to go farther than
that -- when possible. But without a model that is complete, we really
can't draw any conclusions about the system.

But what is _causing_ those changes? The causes could well be outputs
of higher systems of either neural or chemical nature. There's no way
to understand a complex system until every variable has been traced >>to

the output of some system function, or has been determined to be

independent (originating from outside the system). The picture you
present is full of dangling variables.

The picture I presented was a highly simplified one, as it must be in
this forum. The changes I noted have been found to occur as described
both in the intact system and in "reduced" circuity in which the basic
elements are tested in vitro. Thus, while higher systems could be
involved in the intact organism, the mechanism described works as
advertised without them.

I'm not trying to argue with the facts; just trying to get enough of them
to understand the system. Presumably, if these neural connections work
exactly the same way, quantitatively, _in vitro_, then we can stop worrying
about any dependence of these basic reactions on the rest of the system --
they continue to work exactly the same whether they are part of another
system or not. It's like characterizing the response of a muscle to driving
signals -- you're looking at a component of a system and measuring its
input-output function. Whatever relationship this component has with the
rest of the system, the temporal characteristics observed, being invariant,
are simply properties that the rest of the system must cope with, but can't
affect.

But it can still be a significant part of a _system_ of control, >>>yes?

No. Relative to an organized hierarchy, for example, it is like an
independent parasite; its behavior is entirely localized and not
subject to manipulation by the rest of the hierarchy. It's like a
separate organism, not part of a whole system.

Ah, but you are assuming a particular form of organization. There are
other behaviors in which Aplysia contracts its mantle musculature, and
separate neural mechanisms are involved. These mechanisms are
interconnected so as to maintain singleness of action where the >different

behaviors would interfere.

This seems to contradict what you've just been saying. So there ARE
external systems that can influence this local feedback loop? At what point
in the loop do their effects enter? Obviously, _in vitro_ such effects
would be removed, so we wouldn't see the effects of other loops. But _in
vivo_ those other loops would be active. If the other loops can suppress
the response to touches on the mantle, then such suppression could easily
be part of some larger process, even though _in vitro_ something similar is
seen. Also, suppose some other process could _activate_ the response. Then
removal of the activation might leave the default condition you describe.

I'm not trying to be difficult; it's just that I have yet to feel that I've
heard enough to understand what is going on here.

···

------------------------------------

There are ways to put functional hypotheses to test. If I deduced
that eyes are for seeing, I suppose you would argue that this is
poppycock. Perhaps they are there merely to provide support for the
eyelids.

Since they do both, you can't argue that they do NOT provide support
for the eyelids. Eyes provide visual signals, eyes provide support >>for

eyelids, eyes provide signals from an actor to an audience about >>emotions
...

This "functional" stuff is an awkward way to talk about purposive or
goal-driven behavior when there is actually purpose, and it's a way >>of

asserting purposes or goals when there are none. What is the >>"function"
of a ruler? Is it to draw straight lines, to measure >>distances, or to
prop a window open? The function of a ruler is >>whatever function it is
made to serve by the system using it.

However, the functional question asks
what is the role of some particular component within a functioning
system. The little knob on my car radio could serve as a nice hook >from

which to hang a picture, but its function within the radio is to >set the
volume. Within my lowest-level control system, the function of >the muscle
spindles in my biceps muscles is to provide a perceptual >signal (about
muscle length and length rate of change) to the motor >neurons serving that
muscle (and others). We deduce these functions >by examining how they
affect other variables in the system.

So far, you are talking about functions in the way I understand the term:
something that creates a relationship between input variables and output
variables. The volume control creates a relationship between the angle of
the knob and the volume of the sound. The knob is part of a relationship
between your actions and the position of the picture.

In the same way, we can observe what effect
Aplysia's mantle-contraction reflex in response to touch has on >important

variables that impact on Aplysia's ability to survive and >reproduce.

..., but here you depart completely from my usage of the term. Now you're
talking about a purpose that is served by the reflex. This is not "the same
way" at all. Here we don't have one set of variables determining the state
of another variable through a functional relationship. A "reflex" is not a
variable. You've gone up one or more levels of verbal abstraction, and are
now talking about things that require a whole sophisticated human brain
even to perceive. I wish I had a better way to say what I mean here. You
obviously don't see the difference, and I'm not getting what I mean across.

We observe that fish and other aquatic organisms attempt to nibble on
Aplysia's gill. We observe that the resulting damage to the gill >often

leads to Aplysia's death. We observe that rapid contraction of >the gill
into the mantle shields the gill, preventing those organisms >from gaining
access to it and eating it. We infer that mantle >contraction to touch is
a protective reflex (or at least that >protection of the gill is at least
one function of the reflex). An we >can easily test this inference by
disabling the reflex and observing >how this manipulation affects Aplysia's
survival.

This is the _other_ meaning of function, the one I don't believe in. All
that "function" means in this context is "effect." One _effect_ of covering
the gill is to make it inaccessible to nibblers. But what Aplysia is
actually _controlling_ is only the sensation of touch. If controlling this
variable contributes to its survival, then that is lucky for Aplysia, but
Aplysia knows nothing of that. When you speak of "function" in this way,
you have switched from talking about Aplysia's own system to talking about
an external observer's evaluation of the effects of Aplysia's having this
particular organization. You are switching from an Aplysia-centered
viewpoint to an observer-centered viewpoint. The observer-centered
viewpoint includes far more than can be involved in Aplysia's internal
systems.

In this way, I can determine what the reflex does for Aplysia >(function),

without ever asserting anything about teleological >purpose.

This is really the basis of my objection. What a reflex does for Aplysia is
whatever you notice, or conjecture, that it does. Different people might
notice different things. But if we're trying to understand how Aplysia
itself is organized, we have to distinguish between effects of behavior in
general, and _controlled_ effects of behavior -- effects, in other words,
that are actually purposive or teleological. This is the new factor that
PCT introduces into these discussions -- the idea that some of the effects
of Alplysia's behavior are actually intended by Aplysia. Most of the old
ways of talking about these phenomena have been carefully crafted so as to
seem to handle purposive effects without explicitly acknowledging that such
intentions can exist, especially in such lowly creatures as Aplysia.
Because of the general rejection of purpose as a real aspect of behavior,
the language that has developed has no way of distinguishing between
intended and unintended effects of behavior -- effects that are relevant to
the organization of a control hierarchy and those that are only byproducts.
--------------------------------

. . . And by the way, since this is a perfectly
reproducible and predictable phenomenon, no reorganization is involved.

In your own example of reorganization, in which the gain of a control
system is varied so as to establish control, the outcome is also
perfectly reproducible and predictable. So by your logic, it isn't
reorganization either.

Not so. The organization of the system changes at random. The systematic
effect comes about through perceptual selection, and the route to the final
state of the system is unpredictable. In the system you describe there is
no random element; the operation of the system could be described by one
equation with constant parameters.

However, we _are_ observing the proximal _result_ that reorganization >is

supposed to produce in the nervous system: change in synaptic >connectivity.

That applies only to the case where the number of synaptic connections
changes. The rest of the phenomenon could be described by an unchanging
nonlinear equation.

If the reorganization process you envision exists in the human nervous
system, the synaptic changes it produces are likely to be mediated by >the

same or similar biochemical mechanisms as observed in Aplysia's >sensory
synaptic terminals.

Possibly. I wouldn't know. But when you say "mediated" you imply a simple
input-output process, and human reorganization as I see it is certainly not
that: it involves a control loop.

At the very least, the habituation mechanism discovered in Aplysia >shows

that learning (structural changes in the nervous system due to >experience)
can occur in the absence of reorganization; thus >reorganization is not the
only form of learning.

But Aplysia doesn't learn anything, does it? Are there any circumstances
under which this pattern of habituation does NOT occur in exactly the same
way? What you describe is a dynamic input-output function which never
changes its form. How can you call that "learning?"

Whether reorganization as you define it exists at
all remains to be demonstrated.

True.

Best,

Bill P.

[From Bruce Abbott (970716.1800 EST)]

Bill Powers (970716.1212 MDT) --

Bruce Abbott (970716.1015 EST)

Bill, you seem to be doing something to quoted material that screws up the
alignment. Note how the ">" marks end up embedded within the text of the
quoted paragraph:

We are probably still some way from being able to provide the complete
description of the system you demand. However, this need not prevent >us

from examining and learning from the information that has been >brought to
light thus far. We know from observation, for example, >that touching the
siphon brings about contraction of the mantle shelf, >and that this action
diminishes with repeated identical touches of the >siphon, if they occur
frequently enough. We know that this >diminishment of response amplitude
is mediated by changes taking place >in the axon terminals of the sensory
neurons, at their junctions with >the motor neurons that stimulate mantle
contraction _and_ with >interneurons which themselves terminate on those same

motor neurons. We know that the response amplitude diminishes because
less neurotransmitter is released by the sensory neurons with each
impulse, and that this effect is mediated by structural changes,
including inactivation of calcium channels in the axon terminals and >(in

the case of long-term habituation) a reduction in the number of >synaptic
junctions.

If you just block copy the paragraph from the original post and then paste
it into your message using the "paste as quotation" option, it should work
properly. (Don't try to reformat it manually.)

I'm using Eudora Lite with message width set to 80 characters and font set
to Courier New, 9 point, and have experienced no problems with word wrap.

OK, this is the phenomenon. But as I said there are loose variables
dangling -- everywhere you say "because," there is a cause described which
is not described as an effect of something else.

I only said "because" once. I don't know the biochemical mechanism by means
of which the calcium channels become deactivated. The details weren't yet
known as of the time of my reference (1991) and I don't have available to me
a more recent paper in which this may have been spelled out. Keep in mind
that I am only reporting on what I have read; I'm not engaged in this research.

I'm not trying to argue with the facts; just trying to get enough of them
to understand the system. Presumably, if these neural connections work
exactly the same way, quantitatively, _in vitro_, then we can stop worrying
about any dependence of these basic reactions on the rest of the system --
they continue to work exactly the same whether they are part of another
system or not. It's like characterizing the response of a muscle to driving
signals -- you're looking at a component of a system and measuring its
input-output function. Whatever relationship this component has with the
rest of the system, the temporal characteristics observed, being invariant,
are simply properties that the rest of the system must cope with, but can't
affect.

Whatever is going on in this form of habituation in Aplysia appears to be
local to the synaptic terminals of the sensory neurons and dependent on the
pattern of stimulation.

Ah, but you are assuming a particular form of organization. There are
other behaviors in which Aplysia contracts its mantle musculature, and
separate neural mechanisms are involved. These mechanisms are
interconnected so as to maintain singleness of action where the
different behaviors would interfere.

This seems to contradict what you've just been saying. So there ARE
external systems that can influence this local feedback loop? At what point
in the loop do their effects enter? Obviously, _in vitro_ such effects
would be removed, so we wouldn't see the effects of other loops. But _in
vivo_ those other loops would be active. If the other loops can suppress
the response to touches on the mantle, then such suppression could easily
be part of some larger process, even though _in vitro_ something similar is
seen. Also, suppose some other process could _activate_ the response. Then
removal of the activation might leave the default condition you describe.

I've only described the part of the system that relates to habituation.
Habituation as I've described it only occurs when these other inputs are
inactive. The as-yet-undescribed portions have the opposite effect on
response amplitude, so the complete system is capable of altering gain in
either direction.

I'm not trying to be difficult; it's just that I have yet to feel that I've
heard enough to understand what is going on here.

I've tried to present enough of it that the basic principle can be grasped.
Providing a detailed description of the mechanism as thus far elucidated
would take more time and effort that I'm willing to expend. You can find
this information in:

Kandel, E. R. (1991). Cellular mechanisms of learning and the biological
basis of individuality. In Kandel, E. R., Schwartz, J. H., and Jessell, T.
M., _Principles of Neural Science_ (3rd Ed.). Norwalk CT: Appleton & Lange,
P.p. 1009-1031.

In the same way, we can observe what effect
Aplysia's mantle-contraction reflex in response to touch has on
important variables that impact on Aplysia's ability to survive and
reproduce.

..., but here you depart completely from my usage of the term. Now you're
talking about a purpose that is served by the reflex. This is not "the same
way" at all. Here we don't have one set of variables determining the state
of another variable through a functional relationship. A "reflex" is not a
variable. You've gone up one or more levels of verbal abstraction, and are
now talking about things that require a whole sophisticated human brain
even to perceive. I wish I had a better way to say what I mean here. You
obviously don't see the difference, and I'm not getting what I mean across.

I do know what you're trying to get across, but I don't think it applies
here. I'm not talking about a purpose that is served by the reflex; rather
I am still talking about function. I have, however, shifted from talking
about the function of some mechanism within an individual Aplysia's system
to the function of a mechanism within Aplysia in promoting survival across
generations of Aplysia. The mechanism is simply there, and by having it
Aplysia has a better chance of survival, or so the theory goes. Aplysia
itself has no reference for survival, no goal of survival, but its equipment
can be shown empirically to protect it. Better survival is a side-effect of
having evolved a mechanism that contracts the mantle in response to touch.
Presumably this variation has been retained because it does serve this
protective function.

This is the _other_ meaning of function, the one I don't believe in. All
that "function" means in this context is "effect." One _effect_ of covering
the gill is to make it inaccessible to nibblers. But what Aplysia is
actually _controlling_ is only the sensation of touch. If controlling this
variable contributes to its survival, then that is lucky for Aplysia, but
Aplysia knows nothing of that. When you speak of "function" in this way,
you have switched from talking about Aplysia's own system to talking about
an external observer's evaluation of the effects of Aplysia's having this
particular organization. You are switching from an Aplysia-centered
viewpoint to an observer-centered viewpoint. The observer-centered
viewpoint includes far more than can be involved in Aplysia's internal
systems.

Yes, I know. But I'm the one who is trying to understand Aplysia's
structure, not Aplysia. The function of the control system (what it does)
is to keep its controlled variable near the reference despite disturbances
that would push it away. But why should such a control system exist in the
first place? One must ask what is the function of maintaining that variable
near its reference -- what does it do for the organism that possesses such
control? How does it contribute to Aplysia's ability to reproduce its genes?

This is really the basis of my objection. What a reflex does for Aplysia is
whatever you notice, or conjecture, that it does. Different people might
notice different things. But if we're trying to understand how Aplysia
itself is organized, we have to distinguish between effects of behavior in
general, and _controlled_ effects of behavior -- effects, in other words,
that are actually purposive or teleological.

I've already described how conjectures about what a reflex does for Aplysia
can be tested, so it's more than just my opinion versus yours. Like any
scientific hypothesis, it can be evaluated experimentally. If we're trying
to understand how Aplysia itself is organized, we do indeed have to examine
its mechanism and learn how it works. But we also want to understand _why_
those variables we find to be controlled _are_ controlled, why the system is
organized as it is. What is the function, in the life of Aplysia, of
contracting the mantle in response to touch? It's an empirical question.

In your own example of reorganization, in which the gain of a control
system is varied so as to establish control, the outcome is also
perfectly reproducible and predictable. So by your logic, it isn't
reorganization either.

Not so. The organization of the system changes at random. The systematic
effect comes about through perceptual selection, and the route to the final
state of the system is unpredictable. In the system you describe there is
no random element; the operation of the system could be described by one
equation with constant parameters.

We could continue to quibble about this. The only free parameter in your
demo is the system gain. The random selection of gain occurs by magic,
although its frequency of occurrence depends in an entirely predictable way
on the size of persistent error.

But Aplysia doesn't learn anything, does it? Are there any circumstances
under which this pattern of habituation does NOT occur in exactly the same
way? What you describe is a dynamic input-output function which never
changes its form. How can you call that "learning?"

Earlier I defined learning as a change in nervous-system structure resulting
from experience. This is learning by that definition. But keep in mind
that the habituation process is only a part of the picture. A brief shock
to Aplysia's tail (for example) will restore the habituated response to
touch. Any strong stimulus will do so (including a bite to the gill).
Aplysia doesn't know why it does this, but as one trying to understand not
only how, but why, I would suggest that as a side-effect of this change,
Aplysia doesn't just sit there leaving its gill exposed (because of previous
habituation of the reflex) while something devours it.

Regards,

Bruce

[From Bill Powers (970717.0620 MDT)]

Bruce Abbott (970716.1800 EST) --

Bill, you seem to be doing something to quoted material that screws up the
alignment. Note how the ">" marks end up embedded within the text of the
quoted paragraph:

And note what happens to the lines when I use your suggestion of "paste as
quoted." To keep my lines from overlapping when received, I set my edit
window width so it's about half an inch less than the maximum. Since you're
using an 80-column width, your lines wrap in my window. And Eudora Lite 3.0
transmits the lines as they appear in the window, putting a hard return
after every line.

Your quoted material shows the same problem on my screen; I just go through
it and edit it. A nuisance, but this e-mail system doesn't seem to have
been designed for e-mail. The margin problem needs a clever solution which
I don't have.

···

--------------------------------------------------------------------

I don't know the biochemical mechanism by means
of which the calcium channels become deactivated. The details weren't
yet known as of the time of my reference (1991) and I don't have
available to me a more recent paper in which this may have been >spelled

out. Keep in mind that I am only reporting on what I have >read; I'm not
engaged in this research.

If the calcium channels are regulated by a system outside the loop being
considered, the picture is incomplete. I'm not blaming you for the
incompleteness; just noting it.

I've only described the part of the system that relates to >habituation.

Habituation as I've described it only occurs when these >other inputs are
inactive. The as-yet-undescribed portions have the >opposite effect on
response amplitude, so the complete system is >capable of altering gain in
either direction.

This clearly indicates that higher-order control processes are involved in
the intact animal. They have to at least _permit_ habituation. I'll see if
I can obtain the references you cite.

I do know what you're trying to get across, but I don't think it >applies

here. I'm not talking about a purpose that is served by the >reflex;
rather I am still talking about function. I have, however, >shifted from
talking about the function of some mechanism within an >individual
Aplysia's system to the function of a mechanism within >Aplysia in
promoting survival across generations of Aplysia.

This way of talking has developed over the years almost into a formal
system of reasoning -- in fact, I suppose calling it "functional analysis"
means it _is_ a formal system. As a practitioner of this system, you seem
to have found ways of understanding these terms so it doesn't seem to you
that you are saying anything unreasonable.

From my point of view, however, this way of talking sounds downright

animistic. Is there really anything that "promotes survival across
generations?" No matter how you put this idea, it suggests that evolution
is interested in promoting survival -- or that _something_ is. What is
wrong with just sticking to the facts? "The withdrawal of the mantle shelf
has the effect of protecting the gills against injury. Given the presence
of a hostile environment, organisms which behave this way tend to survive
better than those that don't." Does that way of putting it leave out
ANYTHING of importance?

When you speak of "the function" of something, meaning its effect, you
can't help implying that the function was somehow designed to have just
that effect. And when there are multiple effects, you then get into
arguments about what the "true" or "real" function is -- which effect is
the "most important" one. Of course there is no real argument; all such
opinions are subjective; there is no way to get data that show what the
most important effect is.

The mechanism is simply there, and by having it
Aplysia has a better chance of survival, or so the theory goes.

What do you mean by "by having it?" Is there a choice about having it? You
make it sound as though Aplysia was very foresighted to have chosen this
mechanism, in comparison with what would have happened if it had not had
it. "By turning left, he saved half a mile of travel." Doesn't that sound
as if turning left was the means by which he saved half a mile of travel?
In an exactly parallel way, what you have said above is that having this
mechanism was the means by which Aplysia increased its chances of survival.

I think the problem is that you come from a tradition in which NOBODY is
expected to take seriously the idea that organisms behave or do anything as
a means of accomplishing an end. Even though the language of functionalism
mirrors the language of purpose, nobody is expected to take it literally as
implying purposiveness. Thus someone who says "the function of eyes is to
see" is shocked to find that others take this to imply that eyes were
developed to achieve the end of seeing. That would be fine, except that
this same person might say "The function of eating is to fill the stomach,"
and be shocked to find that someone thinks he means that eating is the
means by which the end of filling the stomach is achieved. In this latter
case, of course, the shock is entirely misplaced, because we can show that
filling the stomach is indeed a goal, and eating is indeed the means of
achieving it.

I think this is why functionalists feel free to use what sounds very much
like purposive language even though they do not believe any purposes are
involved. They don't believe in purposive behavior
AT ALL, so they at no risk of being misunderstood by other functionalists.
The result is that they use purposive language for both purposive and
non-purposive processes, without realizing that there is a difference, and
indeed thinking that they are ALWAYS talking about non-purposive processes.

Biologists have long talked like functionalists. A wolf, they say, marks
his territory to warn off other wolves. If you were to ask a biologist who
said that whether wolves really had the intention of warning off other
wolves, and used marking as a means of doing that, the biologist would be
insulted. Of course not! That's just a convenient way of talking; the real
explanation is that natural selection has produced wolves who mark their
territory, and other wolves who tend to avoid entering marked territory.
This is certainly not intentional behavior!

By the same token, a biologist who says that a farmer puts up a "no
trespassing" sign to keep others off his farm is not saying that the farmer
_literally_ uses the sign as a means of carrying out an intention. It's
just that farmers who put up such signs survive better than those who
don't. The biologist thinks that the idea of _really_ intentional or
purposive behavior is hogwash.

People who understand PCT know that there are two kinds of consequences of
actions: intended consequences and unintended consequences. The intended
consequences are controlled by the organism; the unintended ones are
accidents, side-effects. So the PCTer would not use purposive language for
non-purposive effects of action, or vice versa. When you understand that
purposive behavior can actually exist, this distinction becomes
fundamentally important.

I've already described how conjectures about what a reflex does for
Aplysia can be tested, so it's more than just my opinion versus yours.

Yes, you can verify that having this reflex reduces the chance of injury to
the gills. But is this an intended or an unintended effect? Basically all
you are proving is that the reflex has certain consequences, of which the
effect on possible injury is only one. You add nothing by saying that "the
function" of this reflex is "to" reduce the chances of injury, as if no
other consequence mattered, and indeed as if the reflex was deliberately
constructed because it would have this effect. When you use verbal
constructions that are, in other contexts, almost always taken to mean
purposive action, you can't complain when others take it that way.

I'm going to skip the quibbles about what is and is not reorganization.

A brief shock
to Aplysia's tail (for example) will restore the habituated response >to

touch. Any strong stimulus will do so (including a bite to the >gill).

Whoa! So this is _not_ a localized phenomenon in the intact animal. The
calcium channels are evidently part of other control loops, rather than
being the end of an explanatory trail. It is perfectly possible that the
calcium channels are shut down by some other system in the intact animal,
although in vitro they are always shut down (lose their effectiveness upon
stimulation). I think this phenomenon is open to numerous interpretations
other than the "official" one.

Best,

Bill P.

[From Bruce Abbott (970717.1845 EST)]

Bill Powers (970717.0620 MDT)]

Bruce Abbott (970716.1800 EST) --

Bill, you seem to be doing something to quoted material that screws up the
alignment. Note how the ">" marks end up embedded within the text of the
quoted paragraph:

And note what happens to the lines when I use your suggestion of "paste as
quoted."

Yes, it looks fine on my screen. (Really!)

To keep my lines from overlapping when received, I set my edit
window width so it's about half an inch less than the maximum. Since you're
using an 80-column width, your lines wrap in my window.

So set your window width to 80 columns and font to 9 point. That should fix
your wrapping problem. Setting your width to less than 80 forces a word
wrap you don't really want.

And Eudora Lite 3.0
transmits the lines as they appear in the window, putting a hard return
after every line.

In my version the sent text gets a hard return at no more than 76
characters, even though it wraps at 80 (soft return) on my screen.

Your quoted material shows the same problem on my screen; I just go through
it and edit it.

Don't. Reset your screen width to 80 and you shouldn't have to.

I do know what you're trying to get across, but I don't think it >applies

here. I'm not talking about a purpose that is served by the >reflex;
rather I am still talking about function. I have, however, >shifted from
talking about the function of some mechanism within an >individual
Aplysia's system to the function of a mechanism within >Aplysia in
promoting survival across generations of Aplysia.

This way of talking has developed over the years almost into a formal
system of reasoning -- in fact, I suppose calling it "functional analysis"
means it _is_ a formal system. As a practitioner of this system, you seem
to have found ways of understanding these terms so it doesn't seem to you
that you are saying anything unreasonable.

And from my point of view, you seem to have found ways of understanding
these terms so it doesn't seem to you that I am saying anything reasonable!

From my point of view, however, this way of talking sounds downright
animistic. Is there really anything that "promotes survival across
generations?" No matter how you put this idea, it suggests that evolution
is interested in promoting survival -- or that _something_ is. What is
wrong with just sticking to the facts? "The withdrawal of the mantle shelf
has the effect of protecting the gills against injury. Given the presence
of a hostile environment, organisms which behave this way tend to survive
better than those that don't." Does that way of putting it leave out
ANYTHING of importance?

Piffle. This is a non issue. "Organisms which behave this way tend to
survive better than those that don't" and "this behavior promotes survival"
are just alternative ways of saying the same thing, and the latter has the
advantage of economy of words.

When you speak of "the function" of something, meaning its effect, you
can't help implying that the function was somehow designed to have just
that effect.

What you mean is not that I can't help implying, but that you can't help
inferring.

And when there are multiple effects, you then get into
arguments about what the "true" or "real" function is -- which effect is
the "most important" one. Of course there is no real argument; all such
opinions are subjective; there is no way to get data that show what the
most important effect is.

Most parts of living organisms are what they are because previous
generations that happened to posess those features had a reproductive
advantage over those that didn't, so that those features got reproduced.
These features don't have a purpose (no designer intends them to work this
way), but they do have one or more functions in the life of the organism
(that is, in one way or another they tend to improve its reproductive
fitness). My opinions about what those functions may be are subjective, as
is any inference, but they are testable.

The mechanism is simply there, and by having it
Aplysia has a better chance of survival, or so the theory goes.

What do you mean by "by having it?" Is there a choice about having it? You
make it sound as though Aplysia was very foresighted to have chosen this
mechanism, in comparison with what would have happened if it had not had
it. "By turning left, he saved half a mile of travel." Doesn't that sound
as if turning left was the means by which he saved half a mile of travel?
In an exactly parallel way, what you have said above is that having this
mechanism was the means by which Aplysia increased its chances of survival.

You're going to have me pulling my hair out if you keep this up. (Won't
that be interesting?) A means implies an end. The result of the
evolutionary process is that an organism's properties tend to be just those
that enable it to survive in its normal environment. There are beneficial
consequences (or at least there were in the past) to having these
properties, but there is no "end."

Ancestors of today's Aplysia that happened to have inherited a mechanism
that contracts the gill in response to touch survived more often and
reproduced better than those not so equipped. If the contingencies that
resulted in this selection still exist today, then today's version of
Aplysia, which retains this mechanism, will tend to survive better for
having it. It doesn't take much in the way of observation and
experimentation to demonstrate that one important function of this
contraction is protection of the gill from damage, although _Aplysia's_
purpose when contracting the gill is to bring the perception of touch
against the siphon or mantle to some low value.

I think the problem is that you come from a tradition in which NOBODY is
expected to take seriously the idea that organisms behave or do anything as
a means of accomplishing an end. Even though the language of functionalism
mirrors the language of purpose, nobody is expected to take it literally as
implying purposiveness.

First you suggest that I am implying purpose where none exists and now you
suggest that I would deny purpose where it exists. If the language of
functionalism mirrors the language of purpose, it is because the Darwinnian
process has shaped living creatures in such a way that their properties
mirror those of inanimate artifacts whose properties have been chosen by a
purposful designer. The whole _point_ of evolutionary theory is to explain
how such apparently purposeful elements could have arisen from a purposeless
process.

People who understand PCT know that there are two kinds of consequences of
actions: intended consequences and unintended consequences. The intended
consequences are controlled by the organism; the unintended ones are
accidents, side-effects.

They may be side-effects, but many of them are not accidents. Example: Many
animals prefer (control for) sweetness in their food; it is no accident that
by doing so they take in nutritive sugars. If sweetness tended to be
associated with deadly poison instead, it is doubtful that a system
controlling for sweetness would exist in these animals.

So the PCTer would not use purposive language for

non-purposive effects of action, or vice versa. When you understand that
purposive behavior can actually exist, this distinction becomes
fundamentally important.

Aplysia's gill withdrawal reflex is clearly defensive, and I see no danger
in describing it as such. A mechanistic analysis alone can only tell you
what this system controls; it cannot tell you why. This is what a
functional analysis attempts to provide.

A brief shock
to Aplysia's tail (for example) will restore the habituated response >to

touch. Any strong stimulus will do so (including a bite to the >gill).

Whoa! So this is _not_ a localized phenomenon in the intact animal. The
calcium channels are evidently part of other control loops, rather than
being the end of an explanatory trail. It is perfectly possible that the
calcium channels are shut down by some other system in the intact animal,
although in vitro they are always shut down (lose their effectiveness upon
stimulation). I think this phenomenon is open to numerous interpretations
other than the "official" one.

This is a hot area of research; we may have our answer soon. There is a
whole project out there, similar to the human genome project, aimed at fully
describing every neuron in Aplysia and its physiological characteristics,
both alone and as part of the larger systems.

The other mechanism I've alluded to is called "sensitization" and a lot more
is known about it than about habituation. In my example of shock to the
tail, sensory neurons in the tail sysapse with interneurons which in turn
synapse with, of all things, the _presynaptic terminals_ of the same sensory
neurons that sense touch to the siphon. The release of neurotransmitter at
these terminals sets in motion a whole cascade of changes within the siphon
touch neurons, and the effect is opposite to the changes that take place
during habituation; release of neurotransmitter by the siphon neurons is
enhanced; gain increases.

Sorry about the "sluggish" resonse.

Regards,

Bruce

[From Bill Powers (970717.2119 MDT)]

Bruce Abbott (970717.1845 EST)--

So set your window width to 80 columns and font to 9 point. That >should

fix your wrapping problem. Setting your width to less than 80 >forces a
word wrap you don't really want.

Well, (a) I can't set my width to 80 characters, because there's a scroll
bar that takes about 3 of them. And (b) if I could set to 80 characters,
quoting by adding > symbols would still increase every line length by 1,
and produce wraps. Also, other people complain about line lengths greater
than about 70 characters. As to setting my font to 9 points, I'll try it.
How do you count characters in a window of variable width (I mean, without
actually counting them, 1,2,3,...).

···

----------------------------------
As to functionalism, it looks as though there will be no meeting of minds
on this subject, so I give up.

Any progress on analyzing the rat data?

Best,

Bill P.

[From Bruce Abbott (970718.0850 EST)]

Bill Powers (970717.2119 MDT)]

Bruce Abbott (970717.1845 EST)

So set your window width to 80 columns and font to 9 point. That >should

fix your wrapping problem. Setting your width to less than 80 >forces a
word wrap you don't really want.

Well, (a) I can't set my width to 80 characters, because there's a scroll
bar that takes about 3 of them.

You can if you use reduce your font size. Message window width and height
parameters only set the size of the box within which your messages will be
displayed on your screen. Using, e.g., 12 points, there isn't room for 80
characters and the scroll bar, as you note, so long lines wrap. Using 9
points, you can fit 80 characters on a line and still room for a scroll bar
and even a view of Eudora's "desktop" underneath. I don't think these
settings affect where hard returns are placed when the message is
transmitted; in fact, there doesn't seem to be any control for that.

And (b) if I could set to 80 characters,
quoting by adding > symbols would still increase every line length by 1,
and produce wraps.

I haven't found that to be a problem using the settings I suggest.

Also, other people complain about line lengths greater
than about 70 characters.

They might try using a smaller font size, if they have a problem. I haven't
had any complaints about my posts having lines that are too long.

As to setting my font to 9 points, I'll try it.
How do you count characters in a window of variable width (I mean, without
actually counting them, 1,2,3,...).

As far as I know, you can't. I type a line of characters and count them.

----------------------------------
As to functionalism, it looks as though there will be no meeting of minds
on this subject, so I give up.

O.K., but I think you're missing the boat.

Any progress on analyzing the rat data?

I've spent more than a few hours just exploring the data in various ways to
see what emerges. There's a lot to look at. More on that later.

Regards,

Bruce

i.kurtzer (970718)

[From Bruce Abbott (970716.1015 EST

The little knob on my car radio could serve as a nice hook from >which to
hang a picture, but its function within the radio is to set the >volume

...

We deduce these functions by examining how they affect other
variables in the system.
In the same way, we can observe what effect
Aplysia's mantle-contraction reflex in response to touch has on important
variables that impact on Aplysia's ability to survive and >reproduce.

bruce, since their are many consequences of a particular in an
organization how do you say what it REALLY is for? the degree it might be
cast aside without significant differences; the number of times it is
instiated vs. other possible roles it could play; some "optimal" criteria
that results its its greater reproductive success (this is actually a
version of the democratic principle above of sheer number, just
necessarily collected post hoc).
lets say a wing of a bird enters into a snake's mouth;
its such a snug fit and its a relatively common occurance, by a liberal
calculus we might say that its 1/16 of the purpose of wing.

i.

[From Bruce Abbott (970719.0830 EST)]

i.kurtzer (970718) --

bruce, since their are many consequences of a particular in an
organization how do you say what it REALLY is for?

Isaac, your question shows that you weren't paying attention. What is the
perceptual input function FOR? It isn't FOR anything. What does it DO? It
converts an input variable into a neural signal. What is Aplysia's
gill-retraction reflex FOR? It isn't FOR anything. What does Aplysia's
gill-retraction reflex DO? It protects Aplysia's gill. (It may have other
functions as well.) Function has to do with doing, not purpose. Function
can be determined by experiment. Function answers the question, why does
this structure have the particular properties it has rather than some other?

Regards,

Bruce

[From Bill Powers (970719.1323 MDT)]

Bruce Abbott (970719.0830 EST)-- (writing to Isaac Kurtzer)

Isaac, your question shows that you weren't paying attention. What is the
perceptual input function FOR? It isn't FOR anything. What does it DO?
It converts an input variable into a neural signal. What is Aplysia's
gill-retraction reflex FOR? It isn't FOR anything. What does Aplysia's
gill-retraction reflex DO? It protects Aplysia's gill.

I half agree. Asking what something is FOR implies that there is an
intended use of it. Asking what something DOES is much closer to describing
what is observable.

But what Aplysia's gill-retraction reflex DOES is retract the gill,
removing it farther from nibblers and predators. Anything beyond that is a
gratuitous interpretation by the observer. To say it "protects" the gill is
a clear falsehood; whether the result is protection depends on the
persistence and strength of the predator -- if there is one. The gill is
probably retracted far more often when there is no danger than when there is.

Function answers the question, why does
this structure have the particular properties it has rather than some >other?

But this answer has to be fanciful. There is no reason to think that
survival demands only one structure with one set of properties; in fact,
judging from the variety of organisms that live in common environments, I
would say that there are multiple solutions to the problem, all about
equally good, and probably many more than weren't found because the ones
that were found sufficed. Function in this sense is always a _post hoc_
guess. It has no explanatory or predictive power.

Science isn't concerned with "why." That's a theological question. It's
concerned with "how."

Best,

Bill P.

i.kurtzer (970720)

[From Bruce Abbott (970719.0830 EST)]

me:

bruce, since their are many consequences of a particular in an
organization how do you say what it REALLY is for?

b.a:

What is the
perceptual input function FOR? It isn't FOR anything.

o.k.

What does it DO? It
converts an input variable into a neural signal. What is Aplysia's
gill-retraction reflex FOR? It isn't FOR anything. What does Aplysia's
gill-retraction reflex DO? It protects Aplysia's gill.

sure.

Function has to do with doing, not purpose.

great!

Function
can be determined by experiment

greater!!

Function answers the question, why does
this structure have the particular properties it has rather than some

other?

no. This is sneaking in that rascally teleology, that damnable "for", by
reidentifying the "doings" with other considerations (such as some
reason) , thereby, conflating the explanadum and explanans.
Would you agree?
But if you only mean that the organization of the doings=function,
irregardless of theoretic asides then i missed your point and agree with
you.
But now potentially clarified i wonder why i still feel ill-at-ease.
i think you never answered my question because you ultimately said
"experiment" which though rings sweetly to me, but can also have the
"function" of preceding a most sour feeling;
IV-DV experiments when used to explain whole organism purposive behavior
would be one.
Anyway, when we now experiment what would be the criteria for determining
significant functions. Clearly, a concatenation of functions could be
near innumerable, so we must determine the worth of these doings.

example) how do we determine the significant function of the Aplysia at
the focus of whole organismic behaviour? that is what does it do?

i.

[From Bruce Abbott (970721.1210 EST)]

i.kurtzer (970720) --

Bruce Abbott (970719.0830 EST)

Function answers the question, why does
this structure have the particular properties it has rather than some
other?

no. This is sneaking in that rascally teleology, that damnable "for", by
reidentifying the "doings" with other considerations (such as some
reason) , thereby, conflating the explanadum and explanans.
Would you agree?

Again, purpose implies an intended end. Function is about what a mechanism
does.

But if you only mean that the organization of the doings=function,
irregardless of theoretic asides then i missed your point and agree with
you.

The structures of organisms have become what they are through a prolonged
and continuing process in which variations have been retained ("selected")
that tended to promote survival. One can ask how a particular mechanism
contributes to the organism's reproductive fitness. Many mechanisms, for
example, contriubte by stabilizing certain variables, such as blood sugar
level or body temperature, that must be maintained within rather narrow
limits if the organism is to survive. We can examine the mechanism to
answer the quesition, _what_ does it do (e.g., regulates blood sugar level).
But that leaves unanswered the question, _why_ is blood sugar regulated?
Experiment shows that when blood sugar is de-regulated, the organism becomes
sick and may die. As sick or dead organisms do not reproduce their kind as
effectively as healthy, live ones, one can see how keeping blood sugar
tightly within certain limits is adaptive. We have identified one reason
why the organism possesses a control system that keeps blood sugar level
within certain limits.

Notice that this is different from the case of inanimate objects whose
properties have not been shaped by a selective process. One cannot ask why
a rock is hard or has a certain color, although we can determine how it
comes to posess those features. There is no why, because the rock's
properties did not come to be what they are by contributing to the survival
of future generations of rocks.

But now potentially clarified i wonder why i still feel ill-at-ease.
i think you never answered my question because you ultimately said
"experiment" which though rings sweetly to me, but can also have the
"function" of preceding a most sour feeling;
IV-DV experiments when used to explain whole organism purposive behavior
would be one.

I cannot decode the above into anything meaningful. What are you trying to say?

Anyway, when we now experiment what would be the criteria for determining
significant functions. Clearly, a concatenation of functions could be
near innumerable, so we must determine the worth of these doings.

Such determinations must always remain theoretical. A theory as to a
function of some mechanism -- how it contributs to the reproductive fitness
of the organism -- can and must be evaluated as is any scientific
conjecture: through empirical tests. As with any scientific theory, such
conjectures can never be proven correct; the best we can do is to show that
such a theory is consistent with the available evidence.

There is, of course, always the possibility that a particular structure or
mechanism is neutral with respect to the organism's fitness (e.g., blue vs
brown iris color), damaging (as in the case of many genetic defects), or
even has mixed effects depending on context (e.g., inheriting a single gene
"for" sickle-cell anemia confirs resistance to malaria; inheriting two of
them kills). But these things can be discovered through empirical test.

example) how do we determine the significant function of the Aplysia at
the focus of whole organismic behaviour? that is what does it do?

The function of Aplysia, broadly conceived, is to produce more Aplysia.
That is not its purpose (it has no purpose), but that is what it is
organized to do.

Regards,

Bruce

i.kurtzer (970721)

About "doings", "contingencies", and (triple scare quotes !) "functional
explanations" let me reiterate the question differently.
These identifiable doings--my sour feeling following the conclusions
derived from IV-DV techniques applied to behavior; or the twitch of
mollusk gills--what makes these significant for considering single
organisms?..that is, without concern to the red herring of reproductive
success. Why note that the "gill twitches" , instead of "gill displaces
water" , or "gill is eaten by snagglepuss" . The list of its consequences
is certainly large so why focus on one rather than the other. Is their
any principle outside caprice that might justify focusing on these rather
than those as its significant "doings" .

i.

[From Bill Powers (970723.0125 MDT)]

Bruce Abbott (970720.0850 EST)--

You point out the direct, proximal effect while ignoring the less

direct, distal one that in Aplysia's evolutionary history has >"screened

in" this particular adaptation.

For some reason this sentence came floating back to my attention, and I
realized where our problem is. When you speak of a "distal" effect, you're
speaking of a _conceptually_ distal effect, not a _physically_ distal
effect. So you're not talking about physical effects at all -- just layers
of interpretation. Not seeing that clearly, I missed the significance of
your next statement:

I've already indicated that it is an interpretation, that is, a
scientific hypothesis as opposed to a direct observation. The >conditions

that shaped Aplysia lie in its evolutionary history and >therefore are not
subject to direct observation.

This, of course, is the Achilles heel of the theory of natural selection
that (if what you say is true) would relegate it forever to the area of
untestable hypotheses -- and thus _out_ of the area of scientific
hypotheses. Science does not arise from pure reason. An untestable
hypothesis is useless in science.

But what you say is not necessarily true. I forget who did it, but one
research team studied the finches of the Galapagos over many years,
including years of plenty and years of drought. They watched the
populations shift between predominantly large-billed and predominanantly
small-billed several times as the nature of the food supply changed. There
have been other studies, ranging fron moving fish from one pond full of
predators to others free of predators, and the famous Manchester (or was it
Birmingham) moths, and the Cairnes studies of bacteria mutating in response
to changes in nutrient supplies, that demonstrate the same effects.

This makes it clear that it is not the entire evolutionary history of any
organism that shapes it. Populations like modern Aplysia might have existed
many times in the past, with intermediate stages where there were other
related forms. At any point during evolution, there is simply no predicting
what the next stage will be -- the causes of evolution are RANDOM variation
and SELECTIVE retention of characteristics.

If you would consider for a moment my alternative to natural selection, I
think you might find something much closer to what you're talking about.
This view is coming rapidly into the mainstream: the idea that evolution of
forms is an intentional act by organisms (although biologists who are
investigating variable rates of mutation are certainly not calling it that).

In the view of evolution that I propose, the central driving force is
control of the environment in the respects that bear on accuracy of
reproduction. The organism that, by its behavior, can control the variables
on which accuracy of reproduction depends will survive when organisms that
can't will mutate into something else, or go extinct. This criterion is
much more relevant than mere frequency of reproduction. A species that
reproduces rapidly but inaccurately will soon become something different or
will more likely die out -- the sooner for reproducing less accurately.
Rapid reproduction is certainly not an advantage under all circumstances;
if the large-billed finches had reproduced too rapidly, they would quickly
have exhausted the food supply they were best able to eat. What was to the
large-billed finches' advantage was to keep producing more large-billed
finches, but not to reproduce as many as possible. In fact, one of the
adaptations to drought conditions was a decrease in mating frequency.

The precursors to the view I propose are gaining credence as biologists
find more and more instances of variable mutation rates, with the rate
going up when the environment drifts into conditions that are dangerous to
life for a given species. The "E. coli" principle is not yet recognized in
biology, but when it is, the last barrier will be removed to seeing that it
is the organism itself that is in charge of evolution, not the environment.
The environment only proposes; the organism disposes.

This is the crux of the issue between us. Where we differ, it is always
because you consistently pick the conventional view that puts the
environment in control of the organism rather than the control-theoretic
view that puts the organism in control of the (local) environment. This,
indeed, is the only major issue standing between PCT and any conventional
approach to living systems. All else is mere negotiable detail.

It has always been considered unscientific to speak of an organism as if it
had any directing power over anything -- even its own behavior. From what
we know about control systems, we can see that this is a very misguided
prejudice, based entirely on ignorance. This prejudice is the only reason
anyone would accept reinforcement theory, in which aspects of the
environment are given the power to shape behavior. It's the only reason
that evolutionary theory tries to explain the evolutionary process in terms
of environmental forces that shape species.

Inevitably, all attempts to put control in the environment end up
attributing purpose, openly or covertly, to non-living processes. The
language forms in which such purposes are cast are borrowed directly from
the natural-language forms we use to refer to actually purposive behavior.
Normally, when we are asked what some non-living object is "for", we
respond by describing our own, or someone else's, purposes that can be
achieved by _using_ that object. What is a lawnmower for? It is for a
person to use to trim the grass to his own satisfaction -- or for any other
purpose, like propping a door open. When we see a person carrying out some
incomprehensible pattern of actions, we ask "What the hell is he DOING?" --
which means, what purpose of his are his actions accomplishing? We may give
the person a sentence-completion test: "You're doing that to ...?" The
answer we expect is a statement of the purpose of the action -- that
person's purpose. So "for," "do", and "to", in this context, are all part
of the normal means of describing purposive behavior -- except when used by
people who don't believe that purposive behavior exists.

In conventional approaches to living systems, purposive language is often
used, mainly because in many circumstances any other language sounds
artificial and stilted. But lacking any concept of control theory, the
_interpretation_ of this language must always end up putting causes back
into the environment. Thus the "purpose" of a lawnmower becomes inherent in
the lawnmower instead of in its user; we now speak of "the function" of the
lawnmower, meaning not the person's reason for using it, but simply the
physical consequence of its operation (ignoring the fact that it can't
operate itself). Simply by leaving out the purposive agent, often by
omitting the subject of a sentence or phrase, we can make it appear that
the purpose or goal is merely the physical outcome of a physical process:
the ice cream was eaten; taking the job increased his income; pressing the
bar caused food to appear. Purposes and goals become mere consequences. Not
_intended_ consequences, but just consequences. If you don't believe in
purpose, there are no intended consequences.

···

--------------------------------
I know that you understand PCT quite well. But I can't help wondering what
layers of reinterpretation go on in your head where I can't see them. Are
you mentally translating the terms of PCT, like purpose, into non-purposive
equivalents (as you see the equivalence)? Are you making allowances for the
peculiarly unscientific thinking in PCT, and retaining only what can be
turned into the appearance of environmental control?

The reason I ask is that you often seem to feel no qualms about speaking of
the environment as the shaping, guiding, controlling element in behavior.
You said, "The conditions that shaped Aplysia lie in its evolutionary
history ..." without apparently a quiver of discomfort -- whereas I felt a
very LARGE quiver. This is far too parallel with saying "the conditions
that shape the behavior of organisms lie in their history of
reinforcement." Environmental control, pure and simple. Yet we both know
that environments do not control anything; only organisms, or machines they
have built, control.

Perhaps the reason you would speak of evolution this way, but not, perhaps,
the behavior of individual organisms, is that received wisdom still tells
us that evolution is a mechanical process involving nothing more purposive
than survival of the fittest. The logic of natural selection is hard to
deny: organisms that are best equipped to survive survive to reproduce, and
those that don't die out.

But there is an alternative to this view that entails control theory. I
suspect that views similar to mine have been kicking around for a long
time, but because they put the locus of control of evolution inside the
organism, they have been consistently rejected by conventional scientists
-- for that reason and no other. Yet when you think of it, why should
organisms be organized from top to bottom, from their nervous systems to
the operations of genes, as control systems, when this principle does not
apply at all to their changes of form and function on the evolutionary
time-scale? The abrupt shift from internal control to external control
should be enough to arouse our skepticism.
This doesn't say that internal control of evolution is a fact; it just says
that arbitrarily denying it is unjustified. It's no longer as impossible as
scientists have assumed.

When you speak of "the conditions that shaped Aplysia," you are, wittingly
or not, making an assertion that Aplysia did not shape itself. At one time,
that could have been taken for granted and nobody would have raised an
eyebrow. But I am here now, and I am raising an eyebrow, with considerably
more than animism to justify my proposals. Just because I have proposed
what I have proposed, you have to recognize your assertion as also being a
proposition which needs to be justified, not a simple statement of fact.
There is now an alternative to natural selection, an alternative that is
based in the principles of control theory. If you don't take it seriously,
it can only be because you are unwilling to give up this last stand
favoring the idea of environmental control over organisms.

Best,

Bill P.

[From Bruce Abbott (970726.1720 EST)]

Bill Powers (970723.0125 MDT) --

I've been entertaining my parents (up from Florida) this week; now that
they've left I would like to address some of the issues raised in this post.

I've already indicated that it is an interpretation, that is, a
scientific hypothesis as opposed to a direct observation. The
conditions that shaped Aplysia lie in its evolutionary history and
therefore are not subject to direct observation.

This, of course, is the Achilles heel of the theory of natural selection
that (if what you say is true) would relegate it forever to the area of
untestable hypotheses -- and thus _out_ of the area of scientific
hypotheses. Science does not arise from pure reason. An untestable
hypothesis is useless in science.

I agree that if the theory of natural selection were untestable, that fact
"would relegate it forever to the area of untestable hypotheses." However,
there seems to be something missing here, some chain of reasoning that
connects my statement, quoted above, to your conclusion that "if what you
say were true," the theory of natural selection would be untestable. It is
a fact that the conditions that shaped Aplysia lie in the past -- we cannot
be sure that they are identical to those conditions we see operating on
Aplysia today. It does not follow from this that the theory of natural
selection is untestable.

To illustrate in a different context perhaps more familiar to physical
scientists, consider another popular theory: the Big Bang theory of
cosmology. The conditions that existed billions of years ago when the Big
Bang supposedly happened are in the past. Although by using powerful
telescopes we can view conditions much closer to that time than today, it is
not even theoretically possible to view conditions as they were at the time
of the earliest events described by Big Bang theory. By your reasoning, Big
Bang theory therefore must be untestable. Of course, it actually IS
testable (because the Big Bang, if it occurred, has had observable
consequences). For the same reason, so is the theory of evolution.

But what you say is not necessarily true. I forget who did it, but one
research team studied the finches of the Galapagos over many years,
including years of plenty and years of drought. They watched the
populations shift between predominantly large-billed and predominanantly
small-billed several times as the nature of the food supply changed. There
have been other studies, ranging fron moving fish from one pond full of
predators to others free of predators, and the famous Manchester (or was it
Birmingham) moths, and the Cairnes studies of bacteria mutating in response
to changes in nutrient supplies, that demonstrate the same effects.

What I _said_ is necessarily true; what you inferred (incorrectly) from what
I said is demonstrably false.

It was David Lack who studied the finches of the Galapagos; part of this
research was reported years ago in _Scientific American_ in an article
entitled "Darwin's Finches." I am well aware of the evidence you cite and,
like you, see it as supporting the theory of evolution. In other words, we
agree that the theory of evolution is testable, although you have managed to
convey a different impression of my position to your audience.

This makes it clear that it is not the entire evolutionary history of any
organism that shapes it. Populations like modern Aplysia might have existed
many times in the past, with intermediate stages where there were other
related forms. At any point during evolution, there is simply no predicting
what the next stage will be -- the causes of evolution are RANDOM variation
and SELECTIVE retention of characteristics.

Of course. I never suggested otherwise.

If you would consider for a moment my alternative to natural selection, I
think you might find something much closer to what you're talking about.
This view is coming rapidly into the mainstream: the idea that evolution of
forms is an intentional act by organisms (although biologists who are
investigating variable rates of mutation are certainly not calling it that).

In the view of evolution that I propose, the central driving force is
control of the environment in the respects that bear on accuracy of
reproduction. The organism that, by its behavior, can control the variables
on which accuracy of reproduction depends will survive when organisms that
can't will mutate into something else, or go extinct. This criterion is
much more relevant than mere frequency of reproduction. A species that
reproduces rapidly but inaccurately will soon become something different or
will more likely die out -- the sooner for reproducing less accurately.
Rapid reproduction is certainly not an advantage under all circumstances;
if the large-billed finches had reproduced too rapidly, they would quickly
have exhausted the food supply they were best able to eat. What was to the
large-billed finches' advantage was to keep producing more large-billed
finches, but not to reproduce as many as possible. In fact, one of the
adaptations to drought conditions was a decrease in mating frequency.

You may have forgotten, but we discussed this once before, at which time I
told you that I had formulated a similar idea many years ago, when I first
heard about the problem of "punctuated equilibrium." It is, therefore,
hardly an idea I would be opposed to.

This is the crux of the issue between us. Where we differ, it is always
because you consistently pick the conventional view that puts the
environment in control of the organism rather than the control-theoretic
view that puts the organism in control of the (local) environment. This,
indeed, is the only major issue standing between PCT and any conventional
approach to living systems. All else is mere negotiable detail.

I do not deny that there are some important differences in our views, but
your characterization of my position is potentially misleading to readers.
Although you were careful to include the phrase, "where we differ," the
impression is left that I always adopt the "control-by-environment" view.
This is incorrect, as you know: I am quite comfortable with the view that
the organism is in control of the (local) environment. Readers are likely
to infer from your statement that we differ on this, but we do not.
Certainly this must disqualify me as an unbending representative of the
"conventional" view.

The crux of the issue between us has more to do with what I perceive as your
unwillingless to accept the utility of a functional account. You take the
stance (not surprisingly) of the engineer, asking "how is this machine
constructed, and how does it work"? To answer these questions definitively
requires a detailed knowledge of structure, and an ability to measure the
relevant forces and signals.

A functional analysis is designed to provide _some_ useful principles about
the machine's operation _when such information is not available_. If the
mechanism is extremely complex and its signals inaccessible, the mechanistic
approach simply cannot be pursued, at least not until technology provides
the relevant tools. Under these circumstances, a functional analysis allows
some progress to be made.

How? One can observe the _accessible_ variables and determine how they are
related. One does not talk at all about the (unknown) mechanism responsible
for the emergence of these relationships, since such ideas would be entirely
speculative. Such an examination usually reveals consistent, reproducable
phenomena.

Imagine, for example, that you had never heard of radio before and know
nothing about electronics. You are given a radio and now endeavour to learn
as much as you can about it. You fiddle with various knobs and switches
until suddenly, the radio's face lights up and music emerges from the
grilled surface. You turn the knob the other way and the radio reverts to
its former inert state.

At this point you know that the gizmo plays music (perhaps it contains a
miniature orchestra?) and that a particular knob seems to have the function
of turning the music on and off. You turn the radio on again and the music
resumes. However, you discover as you are holding the radio that if you
turn it in a certain direction, the music becomes weaker and buried in a
hissy sound. There seems to be some requirement that the radio be oriented
in a particular direction if the music is to be heard clearly. This doesn't
seem consistent with the idea that the radio contains a miniature orchestra.
A few moments later, the orchestra stops playing and is replaced by a
woman's voice. The voice says that you are listening to radio station WJR
in Detroit and that the time is 8 p.m. You get out a map and determine
where Detroit is relative to your position. If you turn the radio so that
its face is at right angles to a line running from you to Detroit, the
sounds can be heard clearly, but moving the radio to another orientation
weakens the voice and intensifies that hissy sound. You deduce that the
source of the voice and music is probably Detroit and that that the radio
must be "aimed" in the direction of Detriot in order to establish a clear
link with the source.

Nowhere in this investigation have you dealt with mechanism. You are
learning about the functional properties of the mechanism, but not how it is
constructed. However, these functional properties are providing some good
hints about the nature of the mechanism. Any theory that purports to
describe the radio's mechanism must be consistent with these observations:
if the theory does not predict the observed phenomena, it is wrong.
Furthermore, in addition to providing guidance for theory-construction, the
functional account provides a limited but potentially useful set of
operating principles -- you can learn what the radio's various knobs and
dials _do_ and this information allows you to take action when, for example,
the volume is weak. (To improve reception, you can try turning the "volume"
knob up, reorienting the radio, or turning the "tuner" knob slightly.)

Of course, this is imperfect knowledge. If radio's circuitry should break
down, this information will not allow you to repair the radio. Nor will it
allow you to construct a similar mechanism. The functional analysis is only
a first stage in a scientific analysis of the radio; it is incomplete. The
analysis will be complete only when the mechanism has been thoroughly
described and understood. That mechanistic analysis will account for the
observed functional relationships and also show under what conditions those
functions can be expected to hold or fail to hold.

PCT offers a proposed mechanism that, if correct, will account for the
phenomena of behavior, including such phenomena as reinforcement and
extinction, and emergence of particular patterns of behavior under
particular schedules of reinforcement. It will allow one to identify the
conditions under which the "principles of reinforcement" may be expected to
hold and those under which those principles will fail to hold. If PCT
cannot account for these observed phenomena, then as a theory it is
incorrect, and will have to be modified accordingly.

It has always been considered unscientific to speak of an organism as if it
had any directing power over anything -- even its own behavior. From what
we know about control systems, we can see that this is a very misguided
prejudice, based entirely on ignorance. This prejudice is the only reason
anyone would accept reinforcement theory, in which aspects of the
environment are given the power to shape behavior. It's the only reason
that evolutionary theory tries to explain the evolutionary process in terms
of environmental forces that shape species.

It is entirely possible that completely apurposeful mechanisms can and do
produce purposeful ones, i.e., control systems. Thus, one can without
contradiction hold that behavior controls perception and that control
systems emerge through an apurposeful mechanism of random variation and
selective retention. That is currently my position, and you have thus far
failed to provide any convincing evidence that would lead me to a different
conclusion.

Inevitably, all attempts to put control in the environment end up
attributing purpose, openly or covertly, to non-living processes. The
language forms in which such purposes are cast are borrowed directly from
the natural-language forms we use to refer to actually purposive behavior.
Normally, when we are asked what some non-living object is "for", we
respond by describing our own, or someone else's, purposes that can be
achieved by _using_ that object. What is a lawnmower for? It is for a
person to use to trim the grass to his own satisfaction -- or for any other
purpose, like propping a door open. When we see a person carrying out some
incomprehensible pattern of actions, we ask "What the hell is he DOING?" --
which means, what purpose of his are his actions accomplishing? We may give
the person a sentence-completion test: "You're doing that to ...?" The
answer we expect is a statement of the purpose of the action -- that
person's purpose. So "for," "do", and "to", in this context, are all part
of the normal means of describing purposive behavior -- except when used by
people who don't believe that purposive behavior exists.

I can't help it that our language makes it easy to confuse purposeful versus
apurposeful consequences of action, but in my last several posts I have done
my best to distinguish functional analysis from the sort of "back-door"
purposivism you describe. Apparently you find it convenient to deny this
distinction as a tactic aimed at discrediting my position.

You state that "inevitably all attempts to put control in the environment end up
attributing purpose, openly or covertly, to non-living processes." This is
an assertion and not a statement of fact. You then point out that it is
normal to use certain words to convey purpose, but that those "who don't
believe that purposeful behavior exists" would not use these words this way.
Obviously not. How does this support the assertion made at the beginning of
the paragraph? I don't see any connection.

In conventional approaches to living systems, purposive language is often
used, mainly because in many circumstances any other language sounds
artificial and stilted. But lacking any concept of control theory, the
_interpretation_ of this language must always end up putting causes back
into the environment. Thus the "purpose" of a lawnmower becomes inherent in
the lawnmower instead of in its user; we now speak of "the function" of the
lawnmower, meaning not the person's reason for using it, but simply the
physical consequence of its operation (ignoring the fact that it can't
operate itself).

In the case of the lawnmower, we know that its primary function is to mow
grass. Its owner may put it to other uses; thus we may say that the
lawnmower, by virtue of the noise it makes, may function to screen out the
neighbor's rock music. Its owner may use it for that purpose. (Note the
difference between function and purpose in the two sentences preceding this
one.) Owing to its construction, the lawnmower functions as a cutter of
grass. Its owner may or may not use it for that purpose. But we can go
further. We can ask, how does the lawnmower's design serve the
grass-cutting function? Its possession of certain features, such as a sharp
whorling blade positioned a couple of inches above the ground, become
intelligible in terms of this function. We can also identify various parts
of the lawnmower in terms of their contribution to this and other functions.
For example, the wheels permit the lawnmower to be moved forward over the
grass yet to be mowed, at a predetermined height. Whorling blades can be
extremely dangerous to humans; is it a mere coincidence that the blade spins
within a steel enclosure? By examining the lawnmower from the functional
perspective, we can understand many of its design features. The lawnmower
is _adapted_ to its function of cutting grass; however, to the extent that
some of these features endanger its user, other features seem adapted to
protecting the user from accidental harm by the dangerous components.

Simply by leaving out the purposive agent, often by
omitting the subject of a sentence or phrase, we can make it appear that
the purpose or goal is merely the physical outcome of a physical process:
the ice cream was eaten; taking the job increased his income; pressing the
bar caused food to appear. Purposes and goals become mere consequences. Not
_intended_ consequences, but just consequences. If you don't believe in
purpose, there are no intended consequences.

Of course, lawnmowers were designed by people to perform a certain function;
we have a fair idea what criteria had to be satisfied during the design
stage. Living organisms were not designed by anyone, but they have
undergone successive stages of modification in which the "design criterion"
was, according to evolutionary theory, simply the ability to survive long
enough to successfully reproduce. The variations "tried" were not chosen
for test by any intelligent agent, but appeared as the outcomes of
undirected, essentially random processes. These variants, when they
contributed to reproductive fitness, have been selectively retained, and
over generations this process has resulted in a close match between the
requirements for reproductive survival and the "design" features of the
organisms living under those constraints. By examining these features in
light of the requirements of the organism's environment, we can make an
educated guess as to the functions (not purposes) these features serve in
the life of the organism. It would be inappropriate to say that these
features "evolved to serve such-and-such a purpose," because purpose implies
intent (references) and evolution (as traditionally conceived) is an
intentionless process. But it is perfectly rational under this view to ask
what function or functions may be served by a given feature; i.e., how does
this feature contribute to the organism's reproductive fitness within the
environment in which it lives? Not only that, but as I have noted
previously, hypotheses as to function are testable.

Thus, the idea that function is just a back-door way to introduce purpose
does not stand up to analysis. Here is a theoretically apurposeful
mechanism that yields organisms having mechanisms with specifiable functions
in the life of the organism. And among those mechanisms that have emerged
from this purposeless process are those ubiquitous control systems, whose
functions define the very concept of purpose.

When you speak of "the conditions that shaped Aplysia," you are, wittingly
or not, making an assertion that Aplysia did not shape itself.

Wittingly. When you assert that Aplysia shaped itself, you imply that
Aplysia, for example, _intended_ to develop a gill-withdrawal mechanism.
That is of course absurd. If by that you mean that Aplysia possesses a
control system that (a) senses a problem in its ability to keep its
"essential variables" well controlled, and (b) responds by relaxing the
fidelity of its genetic reproductive mechanism, then I am sympathetic but
skeptical. (That is, I see some serious problems with the idea. This is
different from maintaining a closed mind on the issue.)

At one time,
that could have been taken for granted and nobody would have raised an
eyebrow. But I am here now, and I am raising an eyebrow, with considerably
more than animism to justify my proposals. Just because I have proposed
what I have proposed, you have to recognize your assertion as also being a
proposition which needs to be justified, not a simple statement of fact.
There is now an alternative to natural selection, an alternative that is
based in the principles of control theory. If you don't take it seriously,
it can only be because you are unwilling to give up this last stand
favoring the idea of environmental control over organisms.

Given that "control" has a specific meaning in this forum, you must
understand by now that I have no sympathy for the idea of environmental
control (as here defined) over organisms. But environments can impose
constraints without exerting control. A given environment may passively
sift in those organisms which by chance happen to have those features which
together allow them to thrive in that environment, and sift out the rest.
The process has been demonstrated to work, and it does not assert intention
on the part of the environment.

Regards,

Bruce

[From Bill Powes (970726.1809 MDT)]

Bruce Abbott (970726.1720 EST)--

This, of course, is the Achilles heel of the theory of natural
selection that (if what you say is true) would relegate it forever to
the area of untestable hypotheses -- and thus _out_ of the area of
scientific hypotheses.

I agree that if the theory of natural selection were untestable, that
fact "would relegate it forever to the area of untestable hypotheses."
However, there seems to be something missing here, some chain of
reasoning that connects my statement, quoted above, to your conclusion
that "if what you say were true," the theory of natural selection >would

be untestable.

My fault for using dangling references. "What you say" was intended to
refer to your statement that the theory is untestable.

It is a fact that the conditions that shaped Aplysia lie in the past -

No, it is a _theory_ that Aplysia was shaped by "conditions" in the past.
That is the theory of natural selection, in which it is asserted that
survival to reproduce is a sufficiently fine sieve to account for
evolutionary phenonema. Even in the observations I cite as experiments with
evolution, there is nothing to indicate that simple survival could have
produced the specificity of adaptation that we see. Of course if nobody
doubts that natural selection is selective enough, then the fact of
survival will be taken as evidence that natural selection "works." But I am
raising doubts, and I'm not the only one to have had them.

-- we cannot
be sure that they are identical to those conditions we see operating >on

Aplysia today. It does not follow from this that the theory of >natural
selection is untestable.

As I say, if survival is taken to be adequate proof that natural selection
is the only factor at work, then the test is really no test. Your test
shows that with complete loss of this response to touches, Aplysia's
mortality rate increases. This does not prove that this final hookup arose
as a sudden acquisition of this ability from scratch.

To have a real test, you need an alternative explanation to offer a choice
of theories. Mine entails the prediction that when under environmental
stress, the species will begin to mutate more rapidly, showing that the
organisms are playing an active role in the process of change (when I
devised this theory, I did not know that this is true). Being nibbled upon
is not necessarily fatal, particularly if the present-day hookup is only a
slight modification of some previous way in which Aplysia avoided nibbles
on its gills. An injured Aplysia could still reproduce, but under my
proposal it would be much more likely to mutate before reproduction, and if
the offspring still were subject to injury, they too would be more likely
to mutate before reproducing. On the other hand, if the mutation does
result in actual protection against nibbles, the offspring once again
become perfect replicas of the parents. The changes cease and this version
of Aplysia continues to propagate.

Natural selection assumes a constant or randomly variable mutation rate,
not one that is varied by the organism in any particular relation to
environmental stress. There is mounting evidence that organisms contain
mechanisms that affect mutation rate (or rate of change in inherited
characteristics) and that mutation rate rises under stress in a large
number of species. So evolution-as-reorganization does have some supporting
evidence.

To illustrate in a different context perhaps more familiar to physical
scientists, consider another popular theory: the Big Bang theory of
cosmology. The conditions that existed billions of years ago when the
Big Bang supposedly happened are in the past. Although by using >powerful

telescopes we can view conditions much closer to that time >than today, it
is not even theoretically possible to view conditions >as they were at the
time of the earliest events described by Big Bang >theory. By your
reasoning, Big Bang theory therefore must be >untestable. Of course, it
actually IS testable (because the Big Bang, >if it occurred, has had
observable consequences). For the same >reason, so is the theory of
evolution.

I love the way you use "for the same reason." The Big Bang rests on an
extensive and elaborate theory of interactions between energy and matter,
which has been tested in enormous detail in the laboratory for many
decades. The evidence for the Big Bang takes many forms, each tied to a
specific observation such as red shift, stellar populations, morphology,
particle physics, and general relativistic effects -- each with its own
experimental backing. Furthermore, we can observe the universe -- now -- as
it was long ago, because of the finite speed of light.

The theory of natural selection (not of evolution) has one, count 'em one,
variable on which survival depends: fitness, measured as reproductive
success. And I should put "measured" in quotes, because fitness is never
actually measured; it's inferred from survival rate or relative population,
the same phenomenon it is supposed to explain.

It was David Lack who studied the finches of the Galapagos; part of >this

research was reported years ago in _Scientific American_ in an >article
entitled "Darwin's Finches." I am well aware of the evidence >you cite
and, like you, see it as supporting the theory of evolution.

No, it is experimental evidence for the _phenomenon_ of evolution. Given
these modern studies, which don't rely on interpretation of a fossil
record, we can now accept that evolution of forms actually takes place (and
is reversible). The _theory_ that goes with this phenomenon is called
natural selection; it states that only differences in "fitness" are
necessary to account for the changes we see. Lamarkian theory was once
offered as an alternative, and rejected because of experimental evidence
that supposedly ruled out the inheritance of "acquired traits" (although
how losing a tail to an experimenter's knife could be called an acquired
trait still eludes me -- it would have been more appropriate to expose
animals to environmental mutagens, wouldn't it?). And I am proposing
another theory, which is that rate of mutation depends on certain basic
types of error signals, the E. coli principle.

In other words, we agree that the theory of evolution is testable,
although you have managed to convey a different impression of my >position

to your audience.

It is not the theory of evolution, which states that the forms of species
evolve, that needs further test. It is the theory of natural selection, and
the assertion that it is sufficient to account for the changes we see that
has never, in fact, been tested.

If you would consider for a moment my alternative to natural >>selection,

I think you might find something much closer to what >>you're talking about.

You may have forgotten, but we discussed this once before, at which >time

I told you that I had formulated a similar idea many years ago, >when I
first heard about the problem of "punctuated equilibrium." It >is,
therefore, hardly an idea I would be opposed to.

I remember that, but you have apparently abandoned that theory in favor of
the conventional concept of natural selection and external causation.

This is the crux of the issue between us. Where we differ, it is >>always

because you consistently pick the conventional view that puts >>the
environment in control of the organism rather than the control->>theoretic
view that puts the organism in control of the (local) >>environment. This,
indeed, is the only major issue standing between >>PCT and any conventional
approach to living systems. All else is mere >>negotiable detail.

I do not deny that there are some important differences in our views, >but

your characterization of my position is potentially misleading to >readers.
Although you were careful to include the phrase, "where we >differ," the
impression is left that I always adopt the "control-by->environment" view.
This is incorrect, as you know: I am quite >comfortable with the view that
the organism is in control of the >(local) environment. Readers are likely
to infer from your statement >that we differ on this, but we do not.
Certainly this must disqualify >me as an unbending representative of the
"conventional" view.

When I said "where we differ" that is what I meant. In the present
discussion, where we differ is on the matter of environmental versus
organismic control of evolution. You're aware of my proposal on this
matter, and you had even had similar thoughts in the past yourself. Given
the choice, you have elected to go with the conventional view that gives
the environment control.

The crux of the issue between us has more to do with what I perceive >as

your unwillingless to accept the utility of a functional account. >You take
the stance (not surprisingly) of the engineer, asking "how is >this machine
constructed, and how does it work"? To answer these >questions
definitively requires a detailed knowledge of structure, and >an ability to
measure the relevant forces and signals.

Either detailed knowledge, or proposals sufficiently detailed to have
consequences that can be tested.

A functional analysis is designed to provide _some_ useful principles
about the machine's operation _when such information is not >available_.

If the mechanism is extremely complex and its signals >inaccessible, the
mechanistic approach simply cannot be pursued, at >least not until
technology provides the relevant tools. Under these >circumstances, a
functional analysis allows some progress to be made.

A "functional analysis," in engineering, is known as "system analysis." It
depends on building up knowledge about a system in terms of input-output
functions, where now "functions" means specifically mathematical
representations. When the component functions have been well enough
characterized, either by observation or by making theoretical proposals, it
becomes possible to deduce how the entire system will behave. If there are
theoretical components in the system, the deductions about system behavior
then become a test of the theory, and the tests against real behavior
provide a way of modifying the theory.

How? One can observe the _accessible_ variables and determine how >they

are related. One does not talk at all about the (unknown) >mechanism
responsible for the emergence of these relationships, since >such ideas
would be entirely speculative. Such an examination usually >reveals
consistent, reproducable phenomena.

The difficulty here is that the apparent relationships seen in this way are
more often than not misleading. It is essential to speculate about the
parts of the system you can't observe; when you do, you realize that the
same observation could have radically different meanings. Just look at the
"behavioral illusion." Most often, the proximal stimulus is not accessible,
which means that the stimuli you observe are more likely to be disturbances
than controlled variables. When you just take into account the accessible
observations, the stimulus and the response, you are very likely to draw
mistaken conclusions about the organization of behavior, and to overlook
the need to get new kinds of data. It's the speculation that reveals
alternatives to the most obvious interpretation, and that shows what is
missing in the observations. It's the speculation that shows you the need
to look more deeply into a phenomenon instead of just taking it at face value.

Imagine, for example, that you had never heard of radio before and >know

nothing about electronics. You are given a radio and now >endeavour to
learn as much as you can about it. You fiddle with >various knobs and
switches until suddenly, the radio's face lights up >and music emerges from
the grilled surface. You turn the knob the >other way and the radio
reverts to its former inert state.

At this point you know that the gizmo plays music (perhaps it contains >a

miniature orchestra?) and that a particular knob seems to have the

function of turning the music on and off. You turn the radio on again
and the music resumes.

Note that at this point, you have established the function of that knob as
that of turning music on and off (provided that someone turns the knob).
Since you don't understand the radio, you have no conception of turning the
_radio_ on.

However, you discover as you are holding the radio that if you
turn it in a certain direction, the music becomes weaker and buried in >a

hissy sound. There seems to be some requirement that the radio be

oriented in a particular direction if the music is to be heard >clearly.

This doesn't seem consistent with the idea that the radio >contains a
miniature orchestra. A few moments later, the orchestra >stops playing and
is replaced by a woman's voice. The voice says that >you are listening to
radio station WJR in Detroit and that the time is >8 p.m. You get out a
map and determine where Detroit is relative to >your position. If you turn
the radio so that its face is at right >angles to a line running from you
to Detroit, the sounds can be heard >clearly, but moving the radio to
another orientation weakens the voice >and intensifies that hissy sound.
You deduce that the source of the >voice and music is probably Detroit and
that that the radio

must be "aimed" in the direction of Detriot in order to establish a >clear

link with the source.

Nowhere in this investigation have you dealt with mechanism. You are
learning about the functional properties of the mechanism, but not how >it

is constructed. However, these functional properties are providing >some
good hints about the nature of the mechanism. Any theory that >purports to
describe the radio's mechanism must be consistent with >these observations:
if the theory does not predict the observed >phenomena, it is wrong.

All that is quite true, and it's an excellent parable of empirical research
done without any understanding of mechanism. As you say, any theory about
the nature of mechanism has to fit these observations -- most of which, of
course, are redundant and irrelevant once you understand the mechanism.

But this approach always assumes that _someone else_ is going to propose
and discover the mechanisms to explain these rather random observations.
This is what Skinner did. He said to his followers, "Don't worry -- you
don't have to learn any neurology or physiology or physics or chemistry or
mathematics. All you have to do is make good systematic observations and
record them faithfully, so that future generations with better technology
and science can explain them. In fact, even a novelist knows enough to do
these experiments."

And he was right. You don't need any serious technical knowledge to do the
sort of functional analysis you're talking about. Basically all you need is
good language skills and a good imagination. "The function of a giraffe's
long legs and neck is to allow browsing at levels that can't be reached by
other browsing species, thus assuring a continued food supply." Now what,
exactly, did I need to know in order to say that? Is there anything in that
"analysis" that I couldn't have said when I was 12 years old?

What Skinner didn't realize was that we don't have to wait for "future
generations" to explain these phenomena. His assumption was that his own
approach represented the cutting edge of technology; what he couldn't
explain, nobody could explain. But there are plenty of people around NOW
who can explain them, and they were there at the same time Skinner was
building his first device to obtain cumulative records (any engineer, by
the way, would have laughed at that device -- why on earth would you want
to record the _integral_ of the data, smoothing out all the variations that
might tell you something useful?).

Furthermore, in addition to providing guidance for theory->construction,

the functional account provides a limited but >potentially useful set of
operating principles -- you can learn what >the radio's various knobs and
dials _do_ and this information allows >you to take action when, for
example, the volume is weak.

etc.

Why do you think that a person interested in understanding how a radio
works wouldn't run through these elementary manipulations in the first ten
minutes of the examination? What you're describing is merely the
throat-clearing stages of a real investigation of a system -- the easy
part, the first superficial examination. If I handed you an alien artifact
and asked you to analyze it, and you came back with nothing more than the
kind of observation you describe, I would take the project away from you
and hand it to someone with some experience with analyzing systems. I would
expect that kind of data to occupy the first section of the report, after
which the interesting part would commence. We do not live in the age of
Copernicus, when naturalistic observation was, indeed, the best that could
be done to study planetary motions.

PCT offers a proposed mechanism that, if correct, will account for the
phenomena of behavior, including such phenomena as reinforcement and
extinction, and emergence of particular patterns of behavior under
particular schedules of reinforcement. It will allow one to identify >the

conditions under which the "principles of reinforcement" may be >expected
to hold and those under which those principles will fail to >hold. If PCT
cannot account for these observed phenomena, then as a >theory it is
incorrect, and will have to be modified accordingly.

Don't forget that PCT has existed for something like 40 years without ever
having to rely on operant-conditioning phenomena to provide materials for
analysis. You're painting a picture in which radical behaviorism led the
way with its basic observations, making it possible for theories like PCT
to be developed. This, of course, is not at all what happened. PCT, without
relying AT ALL on the observations from operant experiments, has arrived at
explanations of behavior that can be applied to any results of operant
conditioning experiments. The roots of PCT simply don't lie in radical
behaviorism or its findings -- or, for that matter, in the findings of any
conventional branch of psychology. It has grown primarily out of studying
phenomena that the conventional branches of psychology have completely
overlooked: control phenomena.

It is entirely possible that completely apurposeful mechanisms can and >do

produce purposeful ones, i.e., control systems. Thus, one can >without
contradiction hold that behavior controls perception and that >control
systems emerge through an apurposeful mechanism of random >variation and
selective retention. That is currently my position, and >you have thus far
failed to provide any convincing evidence that would >lead me to a
different conclusion.

Convincing evidence isn't the point. What "convincing evidence" is there
for the view you hold? How do you know it's "entirely possible" for natural
selection to come up with purposive systems? What convinced you of that? It
seems to me that this is simply an assumption, with no reason behind it
except that someone proposed this as the way natural selection works. If
you take the first proposal as the preferred one, for no better reason than
that it was proposed first, you have little basis for defending it. Darwin
didn't consider the PCT explanation of evolution and, on the basis of his
observations, reject it in favor of natural selection. Natural selection
was his first and only big idea, and in fact he modeled this concept after
the purposeful selection by animal breeders. He just took it for granted
that there was no "breeder" involved in evolution, and anyway the pigeons
weren't breeding themselves, were they?

I can't help it that our language makes it easy to confuse purposeful
versus apurposeful consequences of action, but in my last several >posts I

have done my best to distinguish functional analysis from the >sort of
"back-door" purposivism you describe. Apparently you find it >convenient
to deny this distinction as a tactic aimed at discrediting >my position.

If that's your best, it's not good enough. It is perfectly possible to say
what you want to say without borrowing from the language of purpose. So why
keep returning to that language, relying on the listener NOT to give it the
usual interpretation? You can speak of the construction and operation of a
lawnmower without saying what it is "for" or what its "purpose" is, or what
its "primary function" is. If you're second-guessing the designer, it's
perfectly OK to speak of the purposes the _designer_ wanted the machine
"for," because then you _are_ speaking of real purposes in a real person's
head. But when you say that a lawnmower is to cut grass, you're putting
function and purpose into the lawnmower that it can't possible have in
isolation from its user. The lawnmower doesn't give a damn about grass --
it will cut anything that's put between its blades. But only if it's pushed.

In the case of the lawnmower, we know that its primary function is to >mow

grass.

That "primary function" does not reside in the lawnmower, but in its user.

<Its owner may put it to other uses; thus we may say that the

lawnmower, by virtue of the noise it makes, may function to screen out
the neighbor's rock music. Its owner may use it for that purpose.

Yes, precisely. The purpose of the lawnmower is its owner's purpose, not
the lawnmower's.

(Note the difference between function and purpose in the two sentences

preceding this one.)

That difference exists only because you fail to see the similarity. There
is no "primary function" inherent in a lawnmower; that could exist only in
the intentions of its designer or purchaser. You can do a complete physical
and engineering analysis of a lawnmower, and nowhere will the analysis
reveal any function, primary or otherwise. If you had never seen a
lawnmover before, you might, at some point, suddenly realize that this
thing could probably cut grass, and that must be what its designer made it
that way to do. But then you're speaking of the designer's purpose, not
"the function" of the machine.

This superficial use of the term "function" gets all mixed up with the
formal meaning of the term. You could, in analyzing the lawnmower, express
the rotational speed of the blades as a function of the forward velocity of
the lawnmower, or the shearing forces on an object as a function of blade
torque and separation of blade from cutter bar. A "function" would be
embodied in the gears and the dimensions of the wheels. But you wouldn't
say that the function of the gears is "to" make the blades rotate -- that's
the purposive way of putting it, and it refers to what the designer wanted
to accomplish. The formal meaning of function simply describes the gears as
creating a constant of proportionality between forward speed and rotational
speed of the cutter blade. The blade speed depends on the forward speed,
and the (linear) function describes the relationship.

What you describe bears the same relationship to real functional analysis
as "positive feedback," meaning "approval", bears to feedback theory.

Owing to its construction, the lawnmower functions as a cutter of
grass.

If it's used that way. It functions as a doorstop if used to hold a door
open. It functions as a weed cutter if it's used to cut weeds. By itself,
for that matter, a lawnmower doesn't "function" at all -- it just sits
there in the shed and rusts until someone gets it out and uses it for
whatever purpose it is used. Your description is neither an objective
mechanical analysis nor a real functional analysis. It's just an exercise
in free association on a commonsense term.

Its owner may or may not use it for that purpose. But we can go
further. We can ask, how does the lawnmower's design serve the
grass-cutting function? Its possession of certain features, such as a
sharp whirling blade positioned a couple of inches above the ground,
become intelligible in terms of this function.

Yes, it becomes intelligible because you can now see a human purpose that
could employ this device as a means of achieving it.

The design didn't just happen. Somebody created it with a purpose in mind.
Somebody called for those blades to be sharp, and specified their position
above the ground. Somebody shaped the blades so they would turn many times
per foot of travel and clip off all the grass between them and the cutter
bar. The lawnmower doesn't just "possess" features; they were deliberately
put there. When you examine the lawnmower and find these features, you're
looking at the motives of the designer; they are what explain how the
lawnmower is constructed, and what suggest its "true function." It's just
like looking at a puzzling computer program, and suddenly saying, "AH, now
I see what he's doing."
"He" is not the program; he's the guy who wrote it.

A lawnmower is just an arrangement of physical components. So is a tree.
What is the "primary function" of a tree? As far as human beings are
concerned, it may be to provide materials for building things out of
lumber. Or to prove nice scenery, or to prove a windbreak. Which of these
is its _true_ primary function? Obviously, none of them. A tree has no
primary function. It is what it is, just as a lawnmower is what it is. Why
can we speak of the primary function of a lawnmower, but not of a tree?
Simply because a human being designed and built the lawnmower to serve a
human purpose, and nobody designed the tree to serve any purpose.

We can also identify various parts
of the lawnmower in terms of their contribution to this and other
functions. For example, the wheels permit the lawnmower to be moved
forward over the grass yet to be mowed, at a predetermined height.
Whorling blades can be extremely dangerous to humans; is it a mere
coincidence that the blade spins within a steel enclosure? By >examining

the lawnmower from the functional perspective, we can >understand many of
its design features.

Can't you see how every sentence in that paragraph is simply am attempt to
deduce the intentions of the designer? It is indeed no coincidence that
there is an enclosure around the blade: someone wanted to protect the user
against injury. That is the _purpose_ or the _function_ of the enclosure.
But the enclosure itself is simply positioned in a way that does not allow
any large objects to come in contact with the blade, or the blade to eject
large objects. You're talking about _why_ it was given those properties. If
there was no designer who positioned it that way, then you can't say which
objects it was designed to exclude and which (like acorns) it excludes or
confines as a side-effect, or whether there was any intention to exclude or
confine anything. Your whole paragraph above is concerned with human
purposes, not abstract and authorless "functions."

The lawnmower is _adapted_ to its function of cutting grass; however, >to

the extent that some of these features endanger its user, other >features
seem adapted to protecting the user from accidental harm by >the dangerous
components.

Dammit, Bruce, using the passive voice only conceals the agent. SOMEONE
ADAPTED the lawnmower to its function of cutting what its user wants cut.
Other features aren't just "adapted" to protecting the user -- they were
installed intentionally to achieve exactly that purpose. Your paragraph
never mentions any agent, human or otherwise, but if you think there was no
human agent just because you avoid mentioning one, you're kidding yourself.
The entire paragraph reeks of purposiveness even though you never mention
the word.

I know just what you're doing here. You're trying to establish a way of
talking about a human artifact that serves a human purpose, but without
mentioning the human being or the purpose. Once you've established that,
you'll say, "In the same way" the components of a living organism "are
adapted" to have certain "primary functions," and now you will be making
the point that blind forces in the environment did the adapting and
established the functions -- that all this arose from natural selection.
And you will insist that you are NOT using purposive language any more,
because now the primary functions just emerge from natural selection.

But this is all a trick. The basic phenomena you start with are purely
purposive; you try to cast them in purposeless terms by using the passive
voice and using verbs without subjects. Then you use this as an entry into
discussing similar phenomena in which there is no human user -- and by
implication, no purposiveness either. But since your parallel is built
entirely on purposive phenomena, it is no parallel at all.

This is by no means a new ploy. Its original purpose was to deny that
evolution was guided or more fundamentally that the world of living
creatures existed to serve God's purposes -- God the Designer. The theory
of natural selection was directly opposed to religious dogma, and every
scientist knew it. And many scientists supported it then, and still do, for
exactly that reason -- it's just another aspect of the war between science
and religion. The resentments engendered by this war have been deep and
bitter, and often violent. The idea of purpose was rejected even in
ordinary behavior for the same reason. Purpose brought us back inevitably
to _God's_ purpose, and scientists were not about to let that back into
their world. Anyone who spoke of purpose was of the Enemy.

What PCT brings into this argument is the concept not of God's purpose but
of the _organism's_ purpose. And not purpose in the sense of the use to
which it could be put by some other agency, or the function it could
fulfill for someone else, but in the sense of its own intentions, its own
ability to specify what it wants to happen to itself, and to act to make
sure it happens. Control theory shows us what no scientist and no
theologian could ever have explained before: what a purpose or intention
is, and how, mechanistically, it works. If God has intentions for living
creatures, then God must have perceptual input functions to tell Him the
current status of these creatures, and reference signals that define their
intended status, and comparators, and error signals that drive the outputs
of reward and punishment that seem to be His primary mode of action. And
His reference signals have to come from somewhere (a joke, friends, a joke).

PCT is not part of the war between science and religion. It suggests a way
in which evolution can be purposive that doesn't automatically bring God
into the picture, or any agency but organisms themselves. The possibility
that organisms themselves might be governed by inner purposes has never
entered the evolutionary debate. Its omission is a serious matter, because
to omit it is to leave out the one mechanism by which evolution could be
efficient, in favor of a process that is driven by abstract rules and
generalizations instead of mechanisms and is of dubious workability.

And now to prove my point:

Of course, lawnmowers were designed by people to perform a certain
function; we have a fair idea what criteria had to be satisfied during
the design stage. Living organisms were not designed by anyone, but >they

have undergone successive stages of modification in which the >"design
criterion" was, according to evolutionary theory, simply the >ability to
survive long enough to successfully reproduce.

There is it, the ploy in which the purposive example is converted to
purposeless form, and then used to show that organisms "have undergone
successive modifications" (by forces in the world around them). The terms
in quotation marks maintain the spurious parallel, by referring to
specifically purposive processes, but with the quotes showing that this is
no longer meant literally.

The variations "tried" were not chosen for test by any intelligent >agent,

but appeared as the outcomes of undirected, essentially random >processes.
These variants, when they contributed to reproductive >fitness, have been
selectively retained, and over generations this >process has resulted in a
close match between the requirements for >reproductive survival and the
"design" features of the organisms >living under those constraints. By
examining these features in

light of the requirements of the organism's environment, we can make >an

educated guess as to the functions (not purposes) these features >serve in
the life of the organism. It would be inappropriate to say >that these
features "evolved to serve such-and-such a purpose," >because purpose
implies intent (references) and evolution (as >traditionally conceived) is
an intentionless process. But it is >perfectly rational under this view to
ask what function or functions >may be served by a given feature; i.e., how
does this feature >contribute to the organism's reproductive fitness within
the

environment in which it lives? Not only that, but as I have noted
previously, hypotheses as to function are testable.

You keep saying they are testable, but what is the test? For Aplysia, it
was to eliminate the ability to withdraw the gill, and demonstrate that
mortality went up (I presume this was an actual experiment). However, this
does not test the idea that Aplysia withdraws its gill to avoid pain, and
that pain is what the species has learned to control. It does not test the
idea that when earlier forms of Aplysia experienced gill damage, they began
to mutate faster, and that the mutation rate slowed again when the new
organization resulted in gill widthdrawal.

Footnote: This hypothesis does not give "fitness" any particular role; it
says that fitness would be a by-product of what is actually being
controlled, since undamaged Aplysia would probably survive to reproduce
better (not that unlimited reproduction is particularly good for the
species; even without predators, the worst enemies of a species are the
members of the same species that require exactly the same resources). But
the test of which you speak would be unable to refute that hypothesis, so
it is no test.

Thus, the idea that function is just a back-door way to introduce >purpose

does not stand up to analysis.

If you still think that you haven't been paying attention to my arguments.
I think I have shown that function, as you use the term, is precisely a
back-door way of talking about phenomena that involve purpose in the PCT
sense.

When you speak of "the conditions that shaped Aplysia," you are,
wittingly or not, making an assertion that Aplysia did not shape itself.

Wittingly. When you assert that Aplysia shaped itself, you imply that
Aplysia, for example, _intended_ to develop a gill-withdrawal >mechanism.

That is of course absurd. If by that you mean that Aplysia >possesses a
control system that (a) senses a problem in its ability to >keep its
"essential variables" well controlled, and (b) responds by >relaxing the
fidelity of its genetic reproductive mechanism, then I am >sympathetic but
skeptical. (That is, I see some serious problems with >the idea. This is
different from maintaining a closed mind on the >issue.)

Good, show me how it's different. Obviously I don't propose that Aplysia
intended to develop a gill-withdrawal mechanism. What are the "serious
problems" with what I do propose? I hope, by the way, that you're not
proposing that natural selection suddenly produced an entire
gill-withdrawal mechanism.

Given that "control" has a specific meaning in this forum, you must
understand by now that I have no sympathy for the idea of >environmental

control (as here defined) over organisms. But >environments can impose
constraints without exerting control. A given >environment may passively
sift in those organisms which by chance >happen to have those features
which together allow them to thrive in that environment, and sift out the
rest.

I have never denied that this phenomenon occurs; it has to occur. And I
have said, and believe, that you understand PCT. It's just that there are
situations where you don't choose to apply it.

The process has been demonstrated to work, and it does not assert
intention on the part of the environment.

The real question is whether it works _well enough_ to account for what we
observe. What I maintain is that only an intentional process could be
efficient enough, and we agree that the non-living environment can't
support any intentional processes. The custom has been to point to
evolutionary phenomena as proof that natural selection is sufficient to
account for the phenomena, and in the absence of any equally plausible
alternative that may have been a reasonable argument. But even without a
control-theoretic alternative, there have been serious questions as to
whether the speed of evolution can be explained by the crude
survival-of-the-fittest principle. The most common way of speaking of
natural selection in this regard is to describe it as remarkably or
surprisingly or astonishingly effective, implying that something tells us
we are asking TOO much of it. But without any plausible alternative, what
can anyone do but say "Well, I guess it must be efficient enough, since it
seems to work?"

I am interested in what you see as serious problems with the control-system
concept of evolution.

Best,

Bill P.

[From Rick Marken (970728.1330 PCT)]

Bill Powers (970726.1809 MDT) to Bruce Abbott (970726.1720 EST) --

A long and wonderfully lucid desciption of the control-theoretic
alternative to the "natural selection" model of the _phenomenon_
of evolution. But perhaps it was so long that Bruce never made it
to the end where Bill says:

I am interested in what you see as serious problems with the
control-system concept of evolution.

This was in response to Bruce's comment:

If...you mean that Aplysia possesses a control system that
(a) senses a problem in its ability to keep its "essential
variables" well controlled, and (b) responds by relaxing the
fidelity of its genetic reproductive mechanism, then I am
sympathetic but skeptical. (That is, I see some serious
problems with the idea.

I would like to know what you see as _serious_ problems with
this proposal, too, Bruce.

Best

Rick

PS. How's the statistics text coming, Bruce? Will it have
a section on how to test for controlled variables?

···

--
Richard S. Marken Phone or Fax: 310 474-0313
Life Learning Associates e-mail: rmarken@earthlink.net
http://home.earthlink.net/~rmarken

[From Bruce Abbott (970729.1025 EST)]

Bill Powers (970726.1809 MDT) --

Yours was an extremely long if hastily composed post, and I spent a fair
amount of time answering it, point by point. I feel much better now, but
I'm not going to send it. I can see the wagons formed into a tight little
circle, and there are rifle barrels pointed at me from behind every wheel.
It is apparent to this lone indian that the settlers are in no mood for
trade. That is too bad, because I have brought some nice jade bracelets and
hand-woven blankets, and you might have had these for free.

Regards,

Bruce

[From Bruce Gregory (970729.1140 EDT)]

Bruce Abbott (970729.1025 EST)]

>Bill Powers (970726.1809 MDT) --

Yours was an extremely long if hastily composed post, and I spent a fair
amount of time answering it, point by point. I feel much better now, but
I'm not going to send it. I can see the wagons formed into a tight little
circle, and there are rifle barrels pointed at me from behind every wheel.
It is apparent to this lone indian that the settlers are in no mood for
trade. That is too bad, because I have brought some nice jade bracelets and
hand-woven blankets, and you might have had these for free.

Digression is the better part of valor.

Bruce