Automatic gain adjustment in Aplysia

[From Bruce Abbott (2004.11.30.1300 EST)]
I’ve been doing a bit of thinking about sea slugs (doesn’t everybody?
(:–> ).
Eric Kandel of Columbia University and others have spent the past couple
of decades studying the physiological mechanisms of learning in the sea
slug, aplysia. For such studies, aplysia offers several
advantages. Its behavior can be modified by experience (that is, it
can learn), its nervous system is relatively simple, its neurons are
large enough to sample the electrical and chemical changes that occur
within them, and the “wiring diagram” of each animal is
identical to that of all the others. Consequently, one can map out the
mechanisms responsible for aplysia’s various behaviors and
identify what changes during learning.
The simplest forms of learning are represented by the processes of
habituation and sensitization, which involve changes in the intensity of
a reflexive response to a given stimulus. In habituation, repeated
presentations of the stimulus at relatively short intervals lead to a
reduction in the strength of the response, perhaps to the point of
disappearance. In sensitization, the reflexive response to a
stimulus becomes stronger after a different, noxious stimulus is
presented elsewhere on the organism’s body.
In aplysia, habituation and sensitization have been investigated
in a defensive reflex. The top of the animal’s body houses a gill
structure and a fleshy siphon within the mantle. Touching the siphon
triggers a rapid contraction of muscles within the mantle walls,
contracting the gill and siphon and covering them with the mantle. After
a few seconds, the muscles slowly relax to expose the gill and siphon
again. Habituation can be demonstrated by touching the siphon every
10 seconds or so. Gradually the muscle contractions become weaker
until the response fails to occur at all. Sensitization can be
demonstrated by delivering a mild shock to the head. If the
gill-withdrawal reflex has not first been habituated, then subsequent
touching of the siphon produces a stronger than normal contraction of the
mantle muscles. If the response has been habituated, delivery of
the shock to the head immediately restores it. (In that case the
sensitization process is referred to as dishabituation.)
Kandel and his colleagues were able to trace out the neurological circuit
responsible for the gill-withdrawal reflex. Sensory neurons that provide
the sense of touch for the siphon synapse with motor neurons in the
abdominal ganglion. There is also a second pathway whereby axon
branches from each sensory neuron synapse with an interneuron, which in
turn synapse with the same motor neuron that receives a direct synaptic
connection from the same sensory neuron. Touching the siphon stimulates
the sensory neurons to produce action potentials, which trigger the
release of a neurotransmitter at the synapses with the interneurons and
motor neurons. The interneurons produce their own action potentials,
triggering the release of neurotransmitter at their synapses with the
motor neurons. The motor neurons fire action potentials, which then
stimulate contraction of the mantle muscles.
So what changes during habituation? It turns out that the amount of
neurotransmitter released from the axon terminals of the sensory neurons
diminishes, thereby reducing the postsynaptic potentials within the
interneurons and motor neurons, making it less likely that the latter
will reach their firing thresholds. Furthermore, the diminished
release of neurotransmitter is due to the inactivation of calcium (Ca++)
channels in the sensory neuron axon terminals. The process whereby
neurotransmitter is released requires an influx of Ca++ ions.
Sensitization of the same reflex occurs because sensory neurons that
detect the shock have axon terminals on the axon terminals of the
sensory neurons of the reflex circuit. Release of neurotransmitter at
these terminals increases the number of calcium channels that will open
when the sensory neurons of the siphon are stimulated, thus increasing
the release of neurotransmitter by these neurons. If the response
had been habituated, this undoes the deactivation of calcium channels
produced during habituation.
In thinking about these processes, it occurred to me that these processes
labeled habituation and sensitization actually represent changes in the
gain of the control system that is attempting to keep the perception of
siphon-touching near zero. Thus, what we have here is a nice
example of a system that automatically adjusts its gain. When the
stimulus is innocuous and repeats regularly, the gain gets turned
down. Under these conditions, whatever is touching the siphon
(perhaps a bit of seaweed waving to an fro in the surf) does not present
any danger to the animal’s delicate gill structure, and the gain can be
safely dialed down to conserve energy and allow respiration to continue.
Painful stimulation arising anywhere would indicate that something
potentially dangerous to the gill may be present and under this condition
a strong protective contraction of the mantle to protect the gill would
be appropriate.
These changes occur as a result of certain experiences and in that sense
represent a simple form of learning, but do not produce a reorganization
of the animal’s nervous system in the sense of adding or subtracting
synapses.
Quite a bit more work has been done with aplysia since Kandel’s
pioneering studies, and researchers today are using this animal to study
the physiological mechanisms of classical conditioning and (even more
recently) instrumental conditioning. Anyone interested in learning
more about this can simply enter “aplysia” and perhaps other
keywords such as “habituation” in a search engine (I prefer
Google) to bring up a treasure trove of information about this
research.

Bruce A.

[From Bill Powers (2004.12.01.0255 MST)]

Bruce Abbott (2004.11.30.1300 EST)–

I’ve been doing a bit of
thinking about sea slugs (doesn’t everybody? (:–>
).

But of course. However, the work by Kandel et. al., as far as I know of
it, leaves a number of questions open that need answers before a model
can be constructed.

Eric Kandel of Columbia
University and others have spent the past couple of decades studying the
physiological mechanisms of learning in the sea slug, aplysia. For
such studies, aplysia offers several advantages. Its
behavior can be modified by experience (that is, it can
learn),

This is too qualitative a statement to be taken at face value. If you are
standing on the deck of a ship in relatively calm water, your muscles
will be making small adjustments to prevent the little tilts of the deck
from disturbing your upright orientation. Now if the waves become larger,
your muscle efforts will change just enough to keep your orientation
upright against the increased disturbance. According to the descriptions
of Aplysia, this could be interpreted to mean that you have “learned
from experience.” You experienced a larger-than-normal disturbance,
and as a consequence your behavior changed. But are we really to call
this “learning”? I would say not. I would call it simply the
normal operation of a control system with unchanged characteristics (your
comments about “no reorganization” indicate that you have a
similar opinion). This would be even more clearly the case if on
repetitions of the experiment we continued to find the same relationship:
small disturbances accompanied by small muscle efforts, large
disturbances accompanied by large muscle efforts.

In the case of Aplysia we seem to have a reaction to a disturbance with
the reaction becoming smaller as the disturbance is prolonged or
repeated, a reaction that becomes immediately larger again when a
“noxious stimulus” is applied – that is, when some other
variable is disturbed. The question that comes to mind is, “Has this
ever occurred before?” And the answer is clearly “yes, it
happens every time.” This does not suggest “learning” to
me. It suggests the operation of a two-level hierarchy of control systems
with constant characteristics.

The simplest forms of learning
are represented by the processes of habituation and sensitization, which
involve changes in the intensity of a reflexive response to a given
stimulus.

This is by no means the only way to interpret the observations. In the
first place, we have to replace “reflexive response to a
stimulus” with “an action that opposes some effect of a
disturbing variable”, where the pressure applied to the siphon is
the disturbing variable and the action is the withdrawal of the siphon
which reduces the amount of pressure.

Suppose we say that there is a higher-level system that can raise
and lower the reference signal for a system that controls touch
sensations from the siphon. This system’s output function is a leaky
integrator. A momentary error signal will make the output increase (which
in this case results in a decrease in the touch reference level), and
the output will then slowly decay toward zero if there are no further
error signals. The decrease in reference level will mean that the
siphon-sensation control system will react to smaller amounts of pressure
by withdrawing the siphon. That decrease will, if no further error
signals occur in the higher system, gradually die out and the reference
signal will rise again to its normal setting, meaning that there will be
less reaction to a touch.

What I’m describing here is the action of a one-way higher-order control
system that controls some variable by means of adjusting the reference
level for another one-way control system of lower order, the system that
keeps sensed pressure on the siphon at or below a given reference level.
As it happens, this reference signal acts on the input function rather
than in a separate comparator – the comparator function is physically
part of the input function.

In habituation, repeated
presentations of the stimulus at relatively short intervals lead to a
reduction in the strength of the response, perhaps to the point of
disappearance. In sensitization, the reflexive response to a
stimulus becomes stronger after a different, noxious stimulus is
presented elsewhere on the organism’s body.

This now begins to suggest a two-way higher system, or a second higher
system that operates to reduce withdrawals of the siphon. As you point
out, withdrawing the siphon poses problems for respiration and locomotion
which are needed for other control processes. The repeated touches that
demonstrate habituation also, by making the siphon withdraw, interfere
with respiration and locomotion, and it would seem likely that other
control systems would act to prevent this by raising the reference signal
for touch – that is, increasing the threshold above which any sensation
of touch would be treated as an error.

Kandel and his colleagues were
able to trace out the neurological circuit responsible for the
gill-withdrawal reflex. Sensory neurons that provide the sense of touch
for the siphon synapse with motor neurons in the abdominal
ganglion. There is also a second pathway whereby axon branches from
each sensory neuron synapse with an interneuron, which in turn synapse
with the same motor neuron that receives a direct synaptic connection
from the same sensory neuron. Touching the siphon stimulates the sensory
neurons to produce action potentials, which trigger the release of a
neurotransmitter at the synapses with the interneurons and motor neurons.
The interneurons produce their own action potentials, triggering the
release of neurotransmitter at their synapses with the motor neurons. The
motor neurons fire action potentials, which then stimulate contraction of
the mantle muscles.

Are these really single-impulse phenomena in the intact animal? The
tendency of neural researchers is to trace single impulses rather than
recording neural frequencies; I’m wondering if that is how Kandel et. al.
did their work too.

Does a touch of a siphon elicit only one spike from one sensory neuron?
Or is there perhaps a train of spikes with a frequency which rises to
some maximum and falls again when the siphon is withdrawn? I’m speaking
of the intact animal here, where the spikes are generated in the normal
way.

This makes a great deal of difference in how we are to understand the
operation of this nervous system. The analysis you report is given in
terms of single neural impulses and jolts of neurotransmitter, but if
these are repetitive processes with variable frequencies we must use a
smoothed representation appropriate to the time-scale of the observed
behavior. Relationships that are visible in that sort of representation
may be totally invisible when one looks only at individual impulses, just
as the operation of a transistor would be incomprehensible if presented
in terms of individual hole and electron behavior.

So what changes during
habituation? … the diminished release of neurotransmitter is due to the
inactivation of calcium (Ca++) channels in the sensory neuron axon
terminals. The process whereby neurotransmitter is released
requires an influx of Ca++ ions.

Sensitization of the same reflex
occurs because sensory neurons that detect the shock have axon terminals
on the axon terminals of the sensory neurons of the reflex
circuit. Release of neurotransmitter at these terminals increases the
number of calcium channels that will open when the sensory neurons of the
siphon are stimulated, thus increasing the release of neurotransmitter by
these neurons. If the response had been habituated, this undoes the
deactivation of calcium channels produced during
habituation.

Missing here is an account of what inactivates the calcium channels. The
obvious answer would seem to be simply that calcium channels are opened
by the higher-order reference signal, and close when that reference
signal diminishes (“diminish” makes sense only if we use
frequency to characterize signals). So the closing of calcium channels is
due to lack of the signal that opens them, the natural state of a calcium
channel evidently being closed. This makes habituation and dishabituation
into simple an decrease or an increase (respectively) of the signals from
the higher system. This would be seen best if the signals were
represented in terms of frequency rather than “occurrances.”

In thinking about these
processes, it occurred to me that these processes labeled habituation and
sensitization actually represent changes in the gain of the control
system that is attempting to keep the perception of siphon-touching near
zero. Thus, what we have here is a nice example of a system that
automatically adjusts its gain.

Exactly what I have been deducing above, so we agree. However, gain
control is not easy to distinguish from control by reference signal in a
nonlinear one-way control system. As I said in B:CP, in a real nervous
system addition has an element of multiplication to it (because of
nonlinearities). In a power-law type of perceptual system, changing a
reference signal not only changes the amount of input that is treated as
zero error, but alters the slope of the error signal as an indicator of
deviations from the reference value. The higher the reference setting,
the lower is the incremental gain. For values near zero, increasing the
reference signal increases the incremental gain because we’re on the toe
of the response curve.

Also, while I agree that it is the perception of touching that is to be
kept near zero, we have to make sure this is interpreted properly. The
perception is the perceptual signal, which is what is actually
controlled. But the pressure corresponding to this perception depends
on the reference signal, because in this case the reference signal is
injected into the input function. In fact, what would seem to be the
perceptual signal is also the error signal, because any excess of
perceptual signal over zero indicates an error condition that is to be
corrected. Changing the reference signal changes the amount of pressure
that corresponds to zero perceptual/error signal.

When the stimulus is
innocuous and repeats regularly, the gain gets turned
down.

Better nail down the antecedents here. “The” stimulus actually
refers to two different stimuli: the noxious ones and the ones applied to
the siphon. The stimuli applied to the siphon are not allowed to become
larger than the amount that results in siphon withdrawal, so they aren’t
normally noxious.

Under these conditions, whatever
is touching the siphon (perhaps a bit of seaweed waving to an fro in the
surf) does not present any danger to the animal’s delicate gill
structure, and the gain can be safely dialed down to conserve energy and
allow respiration to continue. Painful stimulation arising anywhere would
indicate that something potentially dangerous to the gill may be present
and under this condition a strong protective contraction of the mantle to
protect the gill would be appropriate.

Right, but the contraction doesn’t occur simply because it’s appropriate.
It occurs because it controls something that a higher system senses.
There are no open-loop processes in a living organism that are not simply
components of some closed loop. The explanation is not complete until we
have identified the higher-order system.

Are you planning to work up a model of Aplysia?

Best,

Bill P.

[From Kenny Kitzke (2004.12.01)]

<Bill Powers (2004.12.01.0255 MST)>

       Bruce Abbott (2004.11.30.1300 EST)--

I’ve been doing a bit of thinking about sea slugs (doesn’t everybody? (:–> ).

Bruce and Bill: Just a note of encouragement. Thanks for posting something professional, new and highly PCT/HPCT relevant on CSGNet. I thoroughly enjoyed the investigation of the behavior of sea slugs from the classical and PCT psychology perspectives. What a contrast! It is great. And, I believe posts like this is what serious behavorial sciences students will want to read and entice them to join our CSGNet.

This is in sharp contrast to the political reference signals of Rick Marken and the feelings of Marc Abrams. They add next to nothing of value to this forum IMHO.

[From Kenny Kitzke (2004.12.01)]

<Bill Powers (2004.12.01.0255 MST)>

          Bruce Abbott (2004.11.30.1300 EST)--

I've been doing a bit of thinking about sea slugs

(doesn't

everybody? (:--> ).
Bruce and Bill: Just a note of encouragement.

Thanks for posting

somethingprofessional, new and highly PCT/HPCT

relevant on CSGNet.

New you say. NEW? Once more I say: consider,
Robertson & Powers, 1990 Intro to Modern Psychology
pp. 113-119

Best,

Dick R

I thoroughly
enjoyed the investigation of the behavior of sea

slugs from the

classical and PCT
psychology perspectives. What a contrast! It is

great. And, I

believe posts
like this is what serious behavorial sciences

students will want

to read and
entice them to join our CSGNet.

This is in sharp contrast to the political

reference signals of

Rick Marken
and the feelings of Marc Abrams. They add next to

nothing of

···

----- Original Message -----
From: Kenneth Kitzke Value Creation Systems
<KJKitzke@AOL.COM>
Date: Wednesday, December 1, 2004 8:18 am
Subject: Re: Automatic gain adjustment in Aplysia

value to this
forum IMHO.

[From Hank Folson (2004.12.01.0730)]

Bruce Abbott (2004.11.30.1300 EST)

Eric Kandel of Columbia University and others have spent the past couple of decades studying the physiological mechanisms of learning in the sea slug, aplysia. For such studies, aplysia offers several advantages. Its behavior can be modified by experience (that is, it can learn), its nervous system is relatively simple, its neurons are large enough to sample the electrical and chemical changes that occur within them, and the "wiring diagram" of each animal is identical to that of all the others. Consequently, one can map out the mechanisms responsible for aplysia's various behaviors and identify what changes during learning.

This research can be important to PCT for only one of two reasons:
1. It provides strong proof that sea slugs are not control systems, but stimulus-response systems. (I think I detect an S-R approach in the research.)
2. The research provides enough raw data that a control sytem model of the sea slug can be created, as Bill Powers hints in [From Bill Powers (2004.12.01.0255 MST)]. The E. coli research and subsequent PCT computer modeling are an example of what I am talking about.

It would be interesting to approach the authors to determine if their goal is to support a specific theory of psychology/physiology, or to learn all about how sea slugs work.

Sincerely,
Hank Folson

[From Kenny Kitzke (2004.12.1.10:50EST)]

<New you say. NEW? Once more I say: consider,
Robertson & Powers, 1990 Intro to Modern Psychology
pp. 113-119

Best,

Dick R>

Are you saying there is something already in your cited reference regarding the behavior of Aplysia? It was new to me, or am I like a sea slug?

Good hearing from you Dick. You have been a tad quiet lately. I was beginning to worry something like a flu-bug got you!

Best wishes,

KK

[From Dick Robertson, 2004. 12.01.1005CST]

···

From: Kenneth Kitzke Value Creation Systems
<KJKitzke@AOL.COM>
Date: Wednesday, December 1, 2004 9:55 am
Subject: Re: Automatic gain adjustment in Aplysia

[From Kenny Kitzke (2004.12.1.10:50EST)]

<New you say. NEW? Once more I say: consider,
Robertson & Powers, 1990 Intro to Modern Psychology
pp. 113-119

Are you saying there is something already in your

cited reference regardingthe behavior of Aplysia?

Yes, that's what I'm saying.

It was new to me, or am I like a sea slug?

Good hearing from you Dick. You have been a tad

quiet lately. I was

beginning to worry something like a flu-bug got you!

No, it's not flu. I seldom get responses to my
posts, so I 've sort of giving up most posting.
It's nice to hear that you read one. Thanks, Kenny.

Best,

Dick

[From Bill Powers (2001.12.01.1022 MST)]

Dick Robertson (2004.12.01) --

New you say. NEW? Once more I say: consider,
Robertson & Powers, 1990 Intro to Modern Psychology
pp. 113-119

You are so right, Dick. I'd forgotten that chapter on learning that you
wrote. The diagram you included makes it clear that we have only part of
the picture. I hope Bruce Abbott tries to get a more complete diagram and
looks into modeling the system.

I see that I kept saying "siphon withdrawal" where I should have said "gill
withdrawal." If there is no effect of the withdrawal on sensory nerve 24,
the loop I imagined is not closed and the lower circuit is not a control
system. However, we need some behavioral data on Aplysia -- it's hard to
guess at that when all we have is a diagram abstracted by someone else.

In the book you speculate about various possible sensory modalities other
than touch. That would be nice to know about. Also, the nature of "noxious
stimuli" is not clear when electric shock is involved. Organisms (other
than electric eels, I suppose) have no sensory system specialized to detect
electric current, so what the experimenter applies as a shock is really a
mix of stimulation of all kinds of sensory organs that will repond to an
electric current, which includes just about all of them. So we have no way
of knowing what controlled variables are disturbed by an electric shock.

Best,

Bill P.

[From Bryan Thalhammer (2004.12.01.1306 CST)]

Thank you Bruce for the post. Your clear post gave me some thoughts and questions.

I agree with Kenny that a sober post about sea slugs seems to be more in context with the kind of forum that CSGnet is, rather than passionate posts about politics and personal matters. However, I do defer to Dick Robertson that these kinds of considerations of aplysia
and other more simple hierarchies have been considered before, by Robertson, Cziko, Powers, and other PCT theorists and comentators.

Two considerations that this discussion brought to me are first, simple hierarchies are both easier and more interesting to study. I would say that simple hierarchies are more interesting than more complex human ones, in that we can see a controlled perception in a more isolated, laboratory view. And that may be the best way to understand control of lower-level control systems. More simple hierarchies have fewer higher-level control systems to muddy the observer’s view as to what they are “doing” and that the level of complexity may factor into the potential for non-lethal development. I recall that earthworms can learn a path to avoid an electric shock or find food in a maze. Aplysia is or is not more complex, I don’t know. But from a PCT point of view, what dertermines/allows for learning or reorganization? How many control systems or how many levels?

I only have my anecdotal story of a tenacious ladybug in one October a few years ago. At about 7pm, I was writing at my PC, and bang, something hit the table like a pebble being thrown. It wasnt a pebble, but the landing of a ladybug. It kept wandering around and around on my note page, searching it seemed, so I thought it might be hungry. I got some boiled chicken, and gave the bug a speck. By golly, the bug found the food and, getting right up close to it, started eating. When I pushed at the bug, it stayed in its position proximal to the food. When i pushed one way, it pushed back, and when I pushed the other way, it still pushed back, but using different bracing against where it was standing. After about 20 min, the bug pulled away from the food, moving more slowly (it was chicken, not turkey, but maybe it was just full), yet it still was meandering a bit. So, figuring that it was still hungry for something else, I brought some fruit flavored yogurt. Again, the bug, once finding the dessert, was smack up against it, not able to be pushed away. After a shorter time, it started walking, but this time, toward a notebook propped up on the sheet. Heading towards the book, it found the spiral, and climbed up halfway. After a bit I noticed that it stopped moving, but it was still within the paper and spiral, I think I saw it belch :slight_smile: . Pushing against it, it pushed back, not wanting to be dislodged from its sleeping nook. By golly that bug was there in the morning, and then it flew off somewhere else in the house. I had ladybugs all winter, feeding them karo syrup on the window ledge, and I released a few into the spring air next March.

Secondly, complex self-systems, such as m. abrams and r. marken, are also interesting to study and converse with, because we see amazing combinations of the words they choose to send to CSGnet. Simply, while personally, each of us has different views of their posts, Rick and Marc do present clear observable artifacts of their higher-level perceptions, one whose words are caustic to the CSGnet conversation, the other whose words represent a challenge to more conservative members, thus drawing other hierarchies into the fray, and still others to lurk and observe. The struggle for what words to see on the CSGnet is an interesting study of control and social interaction from a PCT standpoint. So, equally, complex organisms are as interesting as aplysia.

The main point I want to make is that we can bring in liberal and conservative politics into such a forum, but it would be better for all of us to discuss these in a more sober tone than the caustic manner in which some contributors have been using. I think the welcome aspect of Bruce’s note is that sobriety, which leaves me the space to think about aplysia and control in a more beneficial manner. It’s not the content, in my estimation, but the manner of presentation.

–Bry

···

-----Original Message-----
From: Control Systems Group Network (CSGnet) [mailto:CSGNET@listserv.uiuc.edu]

** On Behalf Of** Kenneth Kitzke Value Creation Systems
Sent: Wednesday, December 01, 2004 8:19 AM
To:
CSGNET@listserv.uiuc.edu
Subject: Re: [CSGNET] Automatic gain adjustment in Aplysia

[From Kenny Kitzke (2004.12.01)]

<Bill Powers (2004.12.01.0255 MST)>

         Bruce Abbott (2004.11.30.1300 EST)--
I've been doing a bit of thinking about sea slugs (doesn't everybody? (:--> ).

Bruce and Bill: Just a note of encouragement. Thanks for posting something professional, new and highly PCT/HPCT relevant on CSGNet. I thoroughly enjoyed the investigation of the behavior of sea slugs from the classical and PCT psychology perspectives. What a contrast! It is great. And, I believe posts like this is what serious behavorial sciences students will want to read and entice them to join our CSGNet.

This is in sharp contrast to the political reference signals of Rick Marken and the feelings of Marc Abrams. They add next to nothing of value to this forum IMHO.

[From Bill Powers (2004.12.01.0255 MST)]

Missing here is an account of what inactivates the calcium channels. The

obvious

answer would seem to be simply that calcium channels are opened by the

higher-order

reference signal, and close when that reference signal diminishes

("diminish" makes

sense only if we use frequency to characterize signals). So the closing of

calcium

channels is due to lack of the signal that opens them, the natural state of

a calcium

channel evidently being closed. This makes habituation and dishabituation

into

simple a decrease or an increase (respectively) of the signals from the

higher

system. This would be seen best if the signals were represented in terms of

frequency

rather than "occurrances."

Also I wanted an explanation for what inactivates the calcium channels. You
have a concise and clear explanation.
Among different URLs where I have studied how perceptual signals are
continued to other neurons, I appreciate:

You propose a very effective perceptual system:

Suppose we say that there is a higher-level system that can raise and lower

the

reference signal for a system that controls touch sensations from the

siphon. This

system's output function is a leaky integrator. A momentary error signal

will make

the output increase (which in this case results in a _decrease_ in the

touch reference level),

and the output will then slowly decay toward zero if there are no further

error signals.

The decrease in reference level will mean that the siphon-sensation control

system

will react to smaller amounts of pressure by withdrawing the siphon. That

decrease will,

if no further error signals occur in the higher system, gradually die out

and the reference

signal will rise again to its normal setting, meaning that there will be

less reaction to a touch.

I now understand there are almost no limits which control systems we find in
our brain. I think they are difficult to specify. Maybe a rule can be:
"Control systems that I can word, may exist. And if it doesn't exist and the
body needs it, it will come (reorganizing)".
If I should make a simplified Aplysia simulation, Could the Reference,
coming from e.g. 3 outputs from a Relationship level be:
IF ((p1,1+p2,1+p3,1 +p4,1)>m;5.0;IF(n<(p1,1+p2,1+p3,1 +p4,1)<m;3.0;1.0)
where m and n indicate different degrees of a perceptual signal at level 2.
What do you say?
I know there is a big difference between a human organism and an Aplysia.
But I asked myself, why didn't Mr. Kandal find an annulospiral sense ending
in the muscles within the mantle walls sending a stretch perceptual signal
to the motor neurone. They are studying the Aplysia different places around
the world, but nobody name the muscle spindles in the actual muscles.
Bjorn

[From Bill Powers (2004.12.02.0659 MST)]

Bjorn Simonsen (2004.12.01.tttt) – (Bjorn, how putting this
heading at the top of your posts? It makes replying and searching easier.
Note that the [ in front of “From” makes it possible to search
only for the originator’s name without finding all the “From”
words in the internet headers. The closing bracket is optional but
satisfies my sense of symmetry)

If I should make a simplified
Aplysia simulation, Could the Reference,

coming from e.g. 3 outputs from a Relationship level be:

IF ((p1,1+p2,1+p3,1 +p4,1)>m;5.0;IF(n<(p1,1+p2,1+p3,1
+p4,1)<m;3.0;1.0)

where m and n indicate different degrees of a perceptual signal at level
2.

What do you say?

In a model, it could be anything you say it is, but I doubt that Aplysia
does logical calculations (logical implications like If…Then, with
logical operators > and <) . Until I see evidence to the contrary
(obtained by someone who would recognize what the issue is), I assume
that frequency of firing is the proper measure of a neural signal, and
that average concentrations of neurotransmitters and messengers are the
proper measures of chemical signals. For methods of simulating
biochemical systems in terms of varying concentrations, see

Hayashi, K. and Sakamoto, N. (1986).
Dynamic analysis of enzyme systems.

New York: Springer-Verlag

I know there is a big
difference between a human organism and an Aplysia.

But I asked myself, why didn’t Mr. Kandal find an annulospiral sense
ending

in the muscles within the mantle walls sending a stretch perceptual
signal

to the motor neurone.

Not all muscles have annulospirtal sensory endings, and not all organisms
have any muscles containing them. I don’t know specifically about
Aplysia, but the chances are that it lacks stretch receptors. Cockroaches
apparently do have them, if I remember right. But maybe what I remember
is that they don’t.

They are studying the Aplysia
different places around

the world, but nobody name the muscle spindles in the actual
muscles.

The reason may be that spindles do not exist in Aplysia. While Aplysia
has been evolving exactly as long as any other modern organism has, it
may never have required the implied degree of motor control to survive
within its own local minimum of intrinsic error (niche).

Best,

Bill P.

···

Bjorn

From [Marc Abrams (2004.12.02.1045)

In a message dated 12/1/2004 9:21:36 AM Eastern Standard Time, KJKitzke@AOL.COM writes:

[From Kenny Kitzke (2004.12.01)]

This is in sharp contrast to the political reference signals of Rick Marken and the feelings of Marc Abrams. They add next to nothing of value to this forum IMHO.

I am a bit puzzled here. I understand that you are often more concerned with how I say something rather than with what I have to say, but I have tried in the past to engage you in conversation with little success.

A recent example was a posting of yours that dealt a bit with diabetes and I responded and corrected some erroneous info you had. I thought that was a terrific subject to talk about because of both of our concerns over it and the control ramifications are both important and interesting

Could you please tell me what you found objectionable in my reply to you then? Or is it something I walk around with for eternity?

There has only been ONE person I have observed in this forum who has NEVER to my knowledge said anything anyone might construe as being objectionable by HOW (that is, in any way offensive) they responded to someone else and that is Bruce Nevin. So how somebody presents what they have to say, including yourself has from time to time been ‘objectionable’ here on CSGnet.

If you don’t like the topics I choose why not tell me about it? I am not sure what purpose your statement serves? Do you feel better after saying it? I suppose you would and I don’t say that either lightly or mockingly. Do you think that statement was not ‘objectionable’ to me? Of course that’s probably the primary reason you wrote it.

I’'m sorry you feel the way you do and I’m sorry you can’t seem to get over the fact that there are others who just do business differently then you do.

Maybe one day that will change for you. If there is any way we can have a dialogue I’d like to know about it. But unfortunately I don’t think you will respond to this post either.

Have a good holiday.

Marc

[From Bill Powers (2004.12.02.0735 MST)]
Bruce Abbott (2004.11.30.1300 EST) –
The account you give of habituation and sensitization is seen differently
in this paper
http://www.brembs.net/learning/aplysia/
The authors (Brembs et. al.) describe what is going on as “classical
conditioning.” The essential part of the description is

···

============================================================================
A light tactile stimulus that would normally not produce any [gill
movement] in the animal, is repeatedly paired with a strong …
stimulus, usually an electric shock. After a few pairings, the animal
shows the [gill movement after] the light tactile stimulus.

You describe the same thing this way:

…repeated presentations of the [siphon] stimulus at relatively short
intervals lead to a reduction in the strength of the response, perhaps to
the point of disappearance. … the reflexive response to a stimulus
becomes stronger after a different …stimulus is presented elsewhere on
the organism’s body.

In both descriptions I have omitted words that carry unjustified
theoretical loadings. The original versions were:
Brembs et. al.:

So what Aplysia behavior could one use to train it? The main focus
on Aplysia neurobiology has been on
classical
conditioning
of the defensive withdrawal reflex. A light tactile
stimulus that would normally not produce any defensive reflex in the
animal, is repeatedly paired with a strong noxious stimulus, usually an
electric shock. After a few pairings, the animal shows the defensive
reflex to the light tactile stimulus.

==============================================================================

The complete content of your description was

==============================================================================

The simplest forms of learning are represented by the processes of
habituation and sensitization, which involve changes in the intensity of
a reflexive response to a given stimulus. In habituation, repeated
presentations of the stimulus at relatively short intervals lead to a
reduction in the strength of the response, perhaps to the point of
disappearance. In sensitization, the reflexive response to a
stimulus becomes stronger after a different, noxious stimulus is
presented elsewhere on the organism’s body.

==============================================================================

The words learning, habituation, sensitization, reflexive, and noxious
were omitted from this description, to obtain what you said about the
actual phenomenon.

You can see that the phenomenon actually described by Bremb is the same
as what you describe, almost; all the loaded words supply is a subtle set
of theoretical assertions disguised as descriptions.

How do these descriptions differ? In Brembs, the action is picked up in
the middle, so to speak, with

“A light tactile stimulus that would normally not produce any [gill
movement] in the animal…”

That corresponds to the stage of the phenomenon in your description where
habituation has occurred, so presumably the siphon has been touched
(perhaps in a previous repetition of the experiment) and any habituation
has not yet worn off.

Brembs concludes

“is repeatedly paired with a strong … stimulus, usually an electric
shock. After a few pairings, the animal shows the [gill movement after]
the light tactile stimulus.”

According to your description, the “pairing” is irrelevant, in
that, as you say.

“the reflexive response to a stimulus becomes stronger after a
different …stimulus is presented elsewhere on the organism’s
body.”

The effect of the signal from the “different stimulus” on the
calcium channels is sufficient to account for the increase in sensitivity
to the siphon stimulus; there is no requirement, apparently, that the
different stimulus be applied prior to the “light” siphon
touch. So the conclusion Brembs draws about classical conditioning is not
supported by the evidence, as far as I have read it.

Terms like “light” and “mild”, by the way, are
unjustifiable when used for stimuli applied to an organism other than
oneself. A “mild” shock to the head might be experienced as
quite different from the same impulse applied to a finger, or to a
testicle.

It would seem that if, like Bremb et. al., you begin with a firm
conviction about what sort of phenomenon you are observing, you simply
don’t do the parts of the experiment that would reveal that the
conviction is wrong. Why test for something you don’t even suspect might
be happening? If Bremb had omitted applying the “paired” siphon
touch when the other stimulus was applied, the same sensitization would
have occurred, if I am reading what you report right.

Brembs says that Aplysia doesn’t do much except eat and reproduce. He
also presents a movie

[http://www.brembs.net/learning/aplysia/biting/index.html

](http://www.brembs.net/learning/aplysia/biting/index.html)in which Aplysia produces an extremely interesting sequence of
actions in which an outer pair of mouth lips sweep closed, then open
slightly while a sort of tongue (“radula”) exudes and folds as
if to grasp a particle, then withdraws. This sequence is perfectly
designed to sweep scattered food particles to the center of the mouth
area where the radula can grasp them and pull them inside. I think that
is a very interesting behavior, possibly the output part of a control
process, and it ought to be modeled.

In fact, the idea that the only thing you can study about the behavior of
Aplysia is the classical conditioning of a “defensive” (what,
Bremb is a mind reader?) “reflex” is not convincing to me. How
does Aplysia locate food and bring it within reach? How does it remove
itself from contact with stimuli that injure it? How does it locate a
mate, close the distance, and do the Right Thing?

In Brembs paper,

[http://www.brembs.net/learning/aplysia/science2002.html

](http://www.brembs.net/learning/aplysia/science2002.html)Brembs describes, among other things, investigating what neuron
produces signals when food is ingested, and then used artificial
stimulation of that nerve as a “reinforcer” to produce feeding
behavior (like that in the movie, I presume). Of course if that neuron
carried signals indicating the presence of food in the gut, then the
signals could be controlled perceptions of successful eating, and
Aplysia’s purpose in eating could be to produce those perceptual signals
(which they are, by my definition in B:CP). The so-called operant
conditioning simply provided a way for feeding behavior to produce that
signal when in fact there was no food ingested – how could Aplysia tell
the difference? What we learn from this is that when Aplysia thinks it
has found food somewhere, and presumably it is not satiated with food, it
will continue browzing or grazing or whatever it is Aplysias do, to
get more. It would be interesting to know what sort of higher
organization tells the feeding control system to keep looking around for
more food instead of galloping off toward greener pastures. That
shouldn’t be hard to find in such a small nervous system.

This was actually my first hit when searching for Aplysia on the Web. How
much more is there to be discovered? Unless everyone is just doing the
same experiments over and over, there could be a lot.

Best,

Bill P.

[From Bryan Thalhammer (2004.12.02.1315 CST)]

From what I remember of my readings to prepare for my research, what we were doing is The Test of the Controlled Variable. Of course, one cannot prove the non-existence of something by its absence, but what are the structures necessary for a simple organism to demonstrate control given such a Test? Of an open or a closed siphon? Proximity to some target? Position in X-Y-Z? Some organisms can only march up or down a chemical gradient by randomizing direction every few seconds. Some organisms can use appendages to maintain a position against digital (finger) pressure.

Am I focusing on the wrong categories or levels with regard to some presumed or predicted control? Or is there an imperceptible or as yet unknown progression from biochemical reactions/bonds to perceptual control?

How does The Test relate to the discussion of Aplysia, since we have brought up S/R [classical] conditioning, training, habituation and sensitization? :slight_smile:

–Bryan

···

-----Original Message-----
From:
Control Systems Group Network (CSGnet) [mailto:CSGNET@listserv.uiuc.edu]** On Behalf Of** Bill Powers
Sent: Thursday, December 02, 2004 10:19 AM
To: CSGNET@listserv.uiuc.edu
Subject: Re: [CSGNET] Automatic gain adjustment in Aplysia

[From Bill Powers (2004.12.02.0735 MST)]

Bruce Abbott (2004.11.30.1300 EST) –

[From Rick Marken (2004.12.02.1315)]

Bill Powers (2004.12.02.0735 MST)--

Brembs says that Aplysia doesn't do much except eat and reproduce. He also
presents a movie

Aplysia Biting Behavior

Very interesting movie. It does, indeed, give some hints about how Aplysia
eats but it certainly makes you wonder how Aplysia reproduces. Or how it
ever gets a date! That thing is serious ugly.

In fact, the idea that the only thing you can study about the behavior of
Aplysia is the classical conditioning of a "defensive" (what, Bremb is a mind
reader?) "reflex" is not convincing to me. How does Aplysia locate food and
bring it within reach? How does it remove itself from contact with stimuli
that injure it? How does it locate a mate, close the distance, and do the
Right Thing?

Sounds like a whole program of research based on an understanding of the
fact that Aplysia is busy controlling various results of its actions.

Best

Rick

···

--
Richard S. Marken
MindReadings.com
Home: 310 474 0313
Cell: 310 729 1400

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

This email message is for the sole use of the intended recipient(s) and
may contain privileged information. Any unauthorized review, use,
disclosure or distribution is prohibited. If you are not the intended
recipient, please contact the sender by reply email and destroy all copies
of the original message.

[From Bruce Abbott (2004.12.02.1610 EST)]

Bruce
Abbott (2004.11.30.1300 EST)–

Bill Powers (2004.12.01.0255 MST)

When
the stimulus is innocuous and repeats regularly, the gain gets turned
down.

Better nail down the antecedents here. “The” stimulus actually
refers to two different stimuli: the noxious ones and the ones applied to
the siphon. The stimuli applied to the siphon are not allowed to become
larger than the amount that results in siphon withdrawal, so they aren’t
normally noxious.

Under these conditions, whatever is
touching the siphon (perhaps a bit of seaweed waving to an fro in the
surf) does not present any danger to the animal’s delicate gill
structure, and the gain can be safely dialed down to conserve energy and
allow respiration to continue. Painful stimulation arising anywhere would
indicate that something potentially dangerous to the gill may be present
and under this condition a strong protective contraction of the mantle to
protect the gill would be appropriate.

Right, but the contraction doesn’t occur simply because it’s appropriate.
It occurs because it controls something that a higher system senses.
There are no open-loop processes in a living organism that are not simply
components of some closed loop. The explanation is not complete until we
have identified the higher-order system.

We’re not in conflict here, but talking about different levels of
analysis: evolutionary versus mechanistic. At the evolutionary level of
analysis, it occurs because, in the evolutionary history of Aplysia, a
system organized in this way proved to be adaptive. In this sense, it
occurs because in the past, such actions have been appropriate, in
the sense of adaptive.
You suggested that what may be changing during habituation of Aplysia’s
gill-withdrawal reflex is not the loop gain but rather the reference
value coming from some higher level. That’s an interesting
possibility that hadn’t occurred to me. But does it fit what we
know (or think we know!) about the mechanism involved? Repeated
elicitation of the reflex over the short term is accompanied by a gradual
weakening of the reaction. Studies have shown that this weakening is not
due to sensory adaptation: sensory neuron activity produced by touching
the siphon skin does not diminish. Nor is the diminished reaction due to
fatigue of muscles. The changes mediating habituation take place in the
axon terminals of the sensory neurons, where they synapse with the
interneurons and motorneurons.
These changes have been observed not only in the abdominal ganglion, but
also in in vitro preparations, where dissected-out sensory neurons
have spontaneously formed synapses with dissected-out motor
neurons. I’m a little hard pressed to locate the hypothesized
higher-level system here that is raising the reference level during the
habituation process.

The reflex seems to be similar to an eyeblink reflex in that it normally
results in full retraction of the gill. I suspect that one does not
normally see a slight retraction just sufficient to reduce or eliminate
contact with the siphon. If so, then it does not behave like a control
system designed to continually adjust its actions so as to keep tactile
input from the siphon at a certain reference value, but rather like a
mechanism designed to protect the delicate gill structures from being
damaged, under the assumption that tactile input above a certain level
may indicate the presence of something potentially damaging. The
relatively slow relaxation that follows contraction makes sense in this
analysis. In other words, it’s a one-way control system with normally
very high gain and slow relaxation of the output.

Bill Powers (2004.12.02.0735
MST)]

The account you give of habituation
and sensitization is seen differently in this paper

http://www.brembs.net/learning/aplysia/

The authors (Brembs et. al.) describe what is going on as “classical
conditioning.” The essential part of the description
is

Classical conditioning and sensitization are different phenomena and I
suspect that Brembs et al. were confusing the two. The difference between
them is that in classical conditioning, the CS stands in a predictive
relation with the US, whereas in sensitization, the two are
independent. A proper classical conditioning experiment will
include a sensitization control in order to evaluate the degree to which
apparent conditioning of the CS is actually not
conditioning but sensitization instead. (In this context, sensitization is sometimes called “pseudoconditioning” because it mimics the effect of conditioning.)

The words learning, habituation, sensitization, reflexive, and noxious were omitted from this description, to obtain what you said about the actual phenomenon.

. . . all the loaded words supply is a subtle set of theoretical assertions disguised as descriptions.

I appreciate your efforts to strip out theoretical assertions, but object that you have gone too far. Habituation and sensitization are not theoretical terms, they are words used to label objective phenomena. It’s convenient to have a word to use instead of having to repeat the description of the observations to which the word refers every time you want to cue the reader or listener as to the topic of conversation. “Reflex” is potentially more troublesome as one might understand it to imply an open-loop, S-R theory but at the observational level it merely defines a type of correlation between certain observable events. “Noxious” is descriptive if defined in terms of its observable effects on the organism rather than as a subjective state, so I don’t have a serious problem with that term either.

In fact, the idea that the only thing you can study about the behavior of Aplysia is the classical conditioning of a “defensive” (what, Bremb is a mind reader?) “reflex” is not convincing to me. How does Aplysia locate food and bring it within reach? How does it remove itself from contact with stimuli that injure it? How does it locate a mate, close the distance, and do the Right Thing?

Brembs et al. didn’t claim what you say they claimed. In fact, the rest of the article goes on to describe research on operant conditioning of feeding behavior. (See operant conditioning of feeding behavior in Aplysia )You may have been mislead by their statement that Aplysia doesn’t do much except feed and reproduce, but I think that this was just an informal observation about what you typically see Aplysia doing and not a scientific catalog of all the behaviors it is capable of exhibiting. In fact, a quick search of the literature on the web reveals that a number of behaviors in Aplysia are being investigated.

This was actually my first hit when searching for Aplysia on the Web. How much more is there to be discovered? Unless everyone is just doing the same experiments over and over, there could be a lot.

It’s become a huge literature and one with which I am only casually familiar at present. In my first post on this subject I only scratched the surface. (For example, I didn’t even bring up the fact that there are at least two kinds of habituation and sensitization that have been observed in Aplysia, based on different underlying processes. The types I introduced are more accurately referred to as “short term” habituation and sensitization; there are “long-term” versions as well.) However, I’m not satisfied with the brief descriptions of behavior given in these reports and I would really like to find some videos that would give a more complete impression of these behaviors and the changes they undergo during exposure to the experimental conditions. I’m thinking it might be useful to contact the labs where this work is being done and find out if any such videos exist.

Bruce A.

[From Bill Powers (2004.12.02.1646 MST)]

Bruce Abbott (2004.12.02.1610 EST) --

You suggested that what may be changing during habituation of Aplysia's
gill-withdrawal reflex is not the loop gain but rather the reference value
coming from some higher level. That's an interesting possibility that
hadn't occurred to me. But does it fit what we know (or think we know!)
about the mechanism involved?

Changing the gain could have the same effect as changing the reference
value, too. If there is a fixed firing threshold T, the perceptual signal P
will be multiplied by a gain factor G (the openness of the calcium
channel?), so the firing will occur at G*P >= T. With high gain a small
signal can cause the firing; as the gain falls, more perceptual signal is
required to produce the firing, so the effect of a lowered gain is the same
as increasing the reference level above which perceptual signal will
produce an error signal.

Repeated elicitation of the reflex over the short term is accompanied by a
gradual weakening of the reaction. Studies have shown that this weakening
is not due to sensory adaptation: sensory neuron activity produced by
touching the siphon skin does not diminish.

That's still OK, since the gain factor is applied where the perceptual
signal affects the next nerve cell. But wait, I realize that I'm making the
assumption that the calcium channel governs the effect of the input signal
on the post-synaptic potential -- is that right?

Anyhow, the slow decline in sensitivity doesn't really sound like a
higher-order effect; it's more like exhausting a local supply of something.
But that's just an uninformed guess.

Nor is the diminished reaction due to fatigue of muscles. The changes
mediating habituation take place in the axon terminals of the sensory
neurons, where they synapse with the interneurons and motorneurons.

So it could still be a slowly-changing effect from higher up. Well, this is
too much guessing on too little data so I'll give up for now.

The idea of contacting the people doing this work is excellent.

Best,

Bill P.