Modeling reorganization

[Martin Taylor 930115 17:10]

Some long time ago, Bill Powers (920709.2000) said: "I'd like to restate

a ground-rule that I try to adhere to in talking about reorganization.

It is that reorganization can't use any facility that develops out of
reorganization. By this I mean that the processes behind reorganization
themselves can't make use of knowledge about the outside world at any
level, from kinesthetic to cognitive. This may turn out to be too
restrictive a demand, but in my mind answers to the basic questions about
reorganization must apply from the beginning of an individual's life,
before the hierarchy has been significantly elaborated. And implicit in
this view is the idea that the reorganizing system operates in the same way
throughout an individual's life.

Noting the caveats, is this still your viewpoint? As I read the above, it
denies the possibility of "learning to learn," or of "developing creativity."
I can see that, in many cases, what seems like learning to learn is better
seen as learning (reorganizing) Perceptual Input Functions (PIFs) that can
be usefully controlled and that contribute to higher PIFs, but isn't there
a reasonable possibility that among the systems that are reorganized is the
reorganizing system itself?

Perhaps it would be dangerous for an organism to reorganize its reorganization
system during its lifetime, since many possible changes could be fatal to
effective reorganization of the main hierarchy (and thus to the organism).
But some changes must occur (if only by mutation), or there would not have
been the necessary genetic diversity to allow evolutionary improvements.

Your statement implies that the whole structure of the reorganising system is
genetically controlled, and you have repeatedly argued that the intrinsic
variable reference levels are solely under genetic control. Is it really
your opinion that nothing environmental can affect either the reorganizing
system structure or its reference levels? What is the nature of acquired
tolerances, such as to arsenic or heroin, or to high altitude (much higher
hemoglobin levels, which suggests that hemoglobin level is an output of
a controller rather than a controlled variable itself)?

I'm not trying to scare up a new thread of argument. I'm taking too much
time with the ones already going. I'm trying to clear up a backlog of
postings that I kept handy with the intention of responding. My mailbox
takes too long to load, with 45 pending messages before we get to the new
ones each day!

Martin

[From Bill Powers (920709.2000)]

Bruce Nevin (920709.0913) --

I haven't responded to one of your (somewhat rare) posts recently. Mostly
just because I agree with you, so with no error signal ... but don't take
that for a blanket endorsement!

I am imagining such a distributed reorganization system. It seems to
parallel the perceptual control hierarchy by virtue of pervading it.

This is a good starting point. As we take off from it, I'd like to restate
a ground-rule that I try to adhere to in talking about reorganization.

It is that reorganization can't use any facility that develops out of
reorganization. By this I mean that the processes behind reorganization
themselves can't make use of knowledge about the outside world at any
level, from kinesthetic to cognitive. This may turn out to be too
restrictive a demand, but in my mind answers to the basic questions about
reorganization must apply from the beginning of an individual's life,
before the hierarchy has been significantly elaborated. And implicit in
this view is the idea that the reorganizing system operates in the same way
throughout an individual's life.

Suggestions of how this might work are in the discussion of the origins
of life. Feedback loops between molecules and between cells does not
go away with the advent of nervous systems implementing control systems
as we usually discuss them. It seems extremely likely that they are
ongoing in parallel with such higher "orders" (as distinct from
"levels") of control.

Agree. With respect to the genetic-level control systems, my favorite
example is the repair enzymes, a product of DNA that continually restores
DNA to conform with some built-in reference pattern. This is just one of
many mechanisms that renders the organism relatively immune to externally-
induced mutations. In my proposals about the origin of life, the same
organization in a much less elaborate form existed from the beginning --
and as you say, it is still here, working away at the most basic
biochemical level. It's a complex control system, perhaps even a
hierarchical one -- but it's not a reorganizing system.

In particular, the kinds of feedback relations
whereby cells take their "voluntary" places in colonies, or within
fungi, or within various orders of plants, or within animals of
increasing complexity, are probably the same kinds of feedback
relations whereby they do so in the embryology and development of
multicellular organisms, and persist in "vegetative" functions that
preserve systemic integrity--what we call intrinsic error.

Excellent thought. PCT, it seems, suggests a whole new approach to
philogeny, cladistics, or what have you. Control supplies a theme that runs
through evolution and in fact shows its direction (Stephen J. Gould would
be horrified). And the principle of reorganization may offer a parallel
theme: the emergence of systematic control from nonsystematic control.
Perhaps repeated at many levels (orders?).

More specifically, it seems likely to me that the cooperating cells
constitutine each ECS can reorganize themselves so as to reduce or
increase something in their environment, such as neuropeptides.

Please elaborate on neuropeptides -- is this a generic term, or a specific
substance? What do they do? What makes them? Are you speaking of
neurotransmitters in general? Are these proteins with specific functions?

That chronic error in an ECS might result in increased production >and/or

release of neuropeptides, with an influence on the cells of >neighboring
ECSs; might result in chemical or even neural signals that >influence the
number, location, or sensitivity of receptor sites >associated with ECSs
that are neurally connected but not physically >close enough to be
"neighboring" in the same sense. And so on.

I've proposed previously that cell differentiation and the "turning off" of
genes could be a simple control process in cells sharing a common
environment. Imagine a set of cells all of which contain a gene that
specifies a reference level for some substance in the shared environment.
Imagine that there is a spread in the natural reference signals. At first,
all cells experience a deficit of that substance, and all therefore begin
creating it. As more and more cells appear through continuing stages of
cell division, the controlled substance (no puns please) will eventually be
brought to the reference level by the action of all the tiny control
systems controlling for a specific concentration of that substance. No one
control system can maintain the required concentration, but many working
independently and in parallel can.

Eventually the concentration will reach the level specified by cells with
the lowest reference settings. As the number of cells increases, that
concentration will exceed those lowest reference settings, and those cells
will cease to produce a contribution to the total concentration.
Equilibrium will occur when there are just enough cells with the highest
reference settings to produce enough of the substance to shut down all
control systems with a lower reference level, and leave a steady-state
population of cells with the highest reference levels just maintaining a
steady concentration of the substance in question. These are one-way
control systems; errors represent only deficits. Therefore there is no
conflict.

Under the usual cause-effect interpretation, the genes responsible for
producing the substance are "programmed" to "turn off" in some cells. The
PCT version of this explanation is that the genes remain as "active" as
ever, but feedback effects from the general concentration of the substance
are higher than the reference level, making the error go negative. As a
negative output is not possible, these control systems are effectively
turned off. There probably isn't any serious difference between this
interpretation and observations -- "repressor" enzymes are known, for
example, which we would interpret simply as perceptual signals with a
negative feedback connection.

My point is that there can appear to be coordinated actions and even
appportionment of functions among systems of the same level without, in
fact, any superordinate coordinating system existing. Your comment above is
on the same track as my thinking.

The existence of non-neural inter-cellular communication and >cooperation

as a mechanism for reorganization does not preclude other >mechanisms for
reorganization operating in parallel.

I agree ... however, I would not call what I just described above
"reorganization." The reason is that it can be accounted for entirely in
terms of the normal operation of control systems of the normal type. At a
given level of organization, cells containing multiple copies of the same
control systems will automatically divide the labor between them: those
with the highest reference levels for a given substance will end up
maintaining a specific concentration of that substance for ALL the cells.
If you think there are parallels between this principle and the
organization of social systems, so do I.

I think we have to use the concept of reorganization sparingly; it can too
easily become a catch-all for unsolved problems of every kind. I would like
to see as much of the growth of the organism as possible, and as much of
the behavioral hierarchy as possible, accounted for by normal interactions
among normal control systems. So I'm not in favor of ...

For example, the input function I of an ECS may reject one candidate
signal i_1 (or reduce its value) because other signals are not present >to

complement i_1 and so the input requirement of I is not met.

This is similar to a suggestion of Martin Taylor's that I also rejected,
and for the same reason. You're proposing a very complex "E"CS, and I
believe we should resist complexities until observations force us into
accepting them because we can see no alternative. We haven't reached that
point yet; we haven't proven that the normal operation of the hierarchy,
and a SIMPLE principle of reorganization, won't solve the problem. And we
may not be ready for such a proof for a very long time (I have no doubt
that it will be forthcoming).

A serious problem is that the neural signals would somehow have to be
interpreted in terms of the meaning they will be given in the new ECS, so
"complementarity" could be detected despite the fact that all neural
signals are basically alike -- just magnitudes.

In order to model what you propose, we would have to show functions in the
model that could detect input signals and judge their complementarity
WITHOUT combining them in the normal way. Then other functions would have
to be provided that convert this judgment about the potential input signals
into a process you call "rejection," which itself might be difficult to
embody in a model.

Even if you could draw such an elaborated diagram of a CCS (complex control
system), you might have problems with stating what a model based on such a
diagram would actually do, using only the rules you have put into it. You
can't just point to the function you want accomplished as proof, until you
can show that the presented model will actually behave in the imagined way.
We understand quite well how an ECS works, given a black box to accomplish
its input function. We can simulate such systems and discover what such a
system will do, even if the input function is too complex to represent
analytically or in any detail. but we know little about more complex
systems.

Most of the basic rules of thumb of PCT and HPCT are based on known and
demonstrated properties of simple control systems. If we start elaborating
on the simplest organization, we must go very slowly and take small steps,
because at every step we have to re-analyze the whole control system to
find out how our changes and additions have changed its basic properties.
Even the most innocuous change could alter the properties we are familiar
with beyond recognition. The only way to handle this is to introduce small
changes and re-do the analysis and simulations each time to find out
whether we have actually made a qualitative change in the system -- whether
we have created something with radically different rules of behavior. This
isn't the sort of thing that can be done every day or every month --
perhaps not even every year.

... what I understand of reorganization is that it is probably not
control either, but rather influence, exerting (strong) selective
"pressure" as the cells of ECSs "in distress" try different changes in
various aspects of their structure and function that they can change >...

Reorganization IS control: it uses, however, a unique kind of primitive
output function, which acts at random but at variable intervals. The result
is to bring a controlled variable to a reference level, the same result we
get from any control system. The list of functions possibly subject to
these random effects,

  Gain
  Weights on various signals in input and output functions
  Location, number, and activity of neuropeptide receptor sites
  Input function "requesting" an imagined signal to complement
    existing input signals--could lead to changed reference
    signals higher up if error is reduced in imagination
  Neural connections with other ECSs

... is a good one, subject to preceding quibbles. What we have yet to
demonstrate is that random variations in such parameters can actually
result in organized semi-permanent systematic control systems.

In fact, none of your concepts amounts to a model yet, but all of them are
good candidates for the primordial soup of concepts from which we will
eventually evolve a more competent model. These are all things we must try
out in simulation.

... can one cell truly control another (in
the same intra-cellular terms in which a cell controls itself)?

Control systems control variables, not things. Your question thus really
asks, can a control system in one cell control a variable inside another
cell? And I think that pretty much answers itself: not if the same variable
is already under control in the other cell. It doesn't seem likely that a
chemical messenger representing a variable inside one cell membrane could
flow freely out of that cell and into a different one to provide a
perception of the controlled variable, nor that an output signal from one
cell could travel equally freely in the other direction.

Does the specialization of cells for cooperating functions have a
parallel in human differences of temperament, talent, etc., as well as >in

educative specialization for social function?

See comment a couple of pages ago.

···

------------------------------------------------------------------------
RE: modeling reorganization

A preliminary report, especially to Martin Taylor who has begun making some
noises about actually doing some joint simulation research on this subject.

I have been playing with reorganization as a way of solving a system of
linear equations:

   y[m] = SUM(a[m,n]*x[n]) where m = n in all cases.

I actually started with the inverse problem: given a set of inputs x1..xn,
and a set of outputs y1..ym, with m = n, find the matrix of coefficients
a[m,n] that will satisfy all m equations. So this is like perceptual
learning.

The basis for reorganization is the sum of squared errors between r[m] and
y[m], where r[m] is the desired set of values of the functions, and y[m] is
the actual set of values for any given set of coefficients a[m,n]. The
vector x[n] is a fixed list of n numbers.

To reproduce the E. coli method, it's necessary to define a direction in m-
dimensional space, using an auxiliary matrix delta[m,n]. The entries in the
delta matrix are changed independently and at random between positive and
negative limits, to create a "tumble." The matrix is normalized so the sum
of its entries, squared, is 1. This makes the entries into direction
cosines in m-dimensional space. A constant times delta[m,n] is added to the
coefficient matrix a[m,n] on each iteration. "Tumbles," however, occur at
intervals determined by the error signal. Between tumbles, the hyperspace
point a[m][n] moves at a constant velocity (actually, this works best if
this velocity depends on the magnitude of error).

Also, instead of using the error (squared) itself, it's necessary to use
the time rate of change of error as the controlled variable. The interval
between reorganizations of the delta matrix is proportional to the negative
time rate of change of the squared error: the more rapidly the error is
decreasing, the greater is the number of iterations before the next
"tumble." When the error increases, there is a "tumble" on every iteration.
So the loop gain is set quite high -- maybe too high.

With 10 equations in 10 variables, the required matrix emerges after
somewhere between 2500 and 10000 iterations, and perhaps 1/10 that many
reorganizations. The RMS error between the target vector r[m] and the
actual-value vector y[m] is then about 0.001 of the initial error; the
numbers r[m] and y[m] agree to one part in 1000 of the maximum or better.
With m and n equal to 50, convergence occurs, but I haven't run it to
completion -- doing so would take days. This is NOT a parallel computer.
Interestingly, the rate of convergence per iteration with 50 dimensions is
not dramatically slower than that with 10 dimensions, given adjustment of
parameters for best performance. It's just that each iteration takes a LOT
longer with 50 than with 10 equations (with 50 equations, the delta[m,n]
matrix involves 2500 random changes per iteration).

I've already learned some things about this method of reorganization. The
best indicator for triggering tumbles is time rate of change of the error.
The variables being randomly altered must be changed not directly, but by
randomly choosing the rate at which they are altered on each iteration. The
delta matrix effectively creates movement in hyperspace at a constant
velocity or a velocity that decreases systematically with error, with only
the direction being altered at random when there is a reorganization event.
I think I can see now that directly varying the output values (in the
a[m,n] matrix) at random would not lead to systematic approach to a
solution, nor would simply using the magnitude of the error rather than its
rate of change. I don't know that for certain, but it seems likely.

I've tried using the squared error, the RMS error, and the mean error as
the basic error measure, and a constant velocity or a velocity that depends
on linear or squared error. Everything tried works, although convergence
rate is affected. Testing is so slow that I haven't really compared the
different possibilities in any useful way, nor have I found any way of
optimizing things like gain and step size. I'm sure that someone with a
better grasp of n-dimensional mathematics and probability than I have could
derive the optimum settings without all this experimentation.

I've also done one test in which complete control systems were used for
each of 10 dimensions. The result converges. But I haven't tested yet with
randomly varying reference signals. Neither have I set up any intrinsic
variables (other than the error signal itself) which are affected variously
by the controlled variables x[n], so that reorganization is based on an
indirect effect of the controlled variables. It turns out that there is an
enormous number of possibilities and variants to investigate; getting to
them all will take some time.

I hope to make a little more progress on this before the meeting. I have to
go to Boulder and Denver next week (a talk on What is Information, a panel
at the meeting of the International Society for Systems Science, into which
I was sweet-talked by Peter Corning -- I'll have a copy of my remarks for
distribution at the meeting and will put it on the net, too, afterward,
with permission from the ISSS). So I won't have a lot of time for this
until later in the summer, after the meeting.

One thing's sure: there's still a lot to learn about this process of
reorganization, even with simple linear systems. And it looks just as
powerful as I thought it would be.
-------------------------------------------------------------------
Best to all,

Bill P.

[Martin Taylor 920710 15:30]
(Bill Powers to Bruce Nevin 920709.2000)

You're proposing a very complex "E"CS, and I
believe we should resist complexities until observations force us into
accepting them because we can see no alternative. We haven't reached that
point yet; we haven't proven that the normal operation of the hierarchy,
and a SIMPLE principle of reorganization, won't solve the problem. And we
may not be ready for such a proof for a very long time (I have no doubt
that it will be forthcoming).

OK. You have a problem with some of my suggestions because they are not
SIMPLE, which is a good reason. It is the same reason I have a problem with
your principle of reorganization using a structure separate from the hierarchy,
and why I tried to replace it with a mechanism inherent in the normal operation
of the hierarchy. I think it is not simple. Simplicity is in the eye of the
beholder, to some extent, but it nevertheless is the only legitimate way of
choosing between theories that claim the same precision and range of
description of observables.

Reading not so far between the lines of your other posting(s), I think you
do not like the idea that intrinsic variables and their reference levels
could be at the top level of the hierarchy, whereas I find that concept to
be both evolutionarily and developmentally natural. If you reject that
concept a priori, then you must see my concept of local reorganization as
more complex than your concept of global reorganization. (I note that you
reluctantly leave the door open for some degree of locality in reorganization,
and you explicitly permit locality of level in reorganization; I allow no such
exceptions in the structure I propose).

Let me reiterate what I am actually proposing, because you seem to have some
slight misunderstanding of it. I start out by denying something you said,
because it contradicts your basic principles, which I embrace:

(920704.0800)

As I see the reorganizing system, it is concerned with controlling INTERNAL
variables only. To reply to a previous comment of yours, I see these
variables as including variables not available to the senses, even to
proprioception.

In my way of looking at things, ANY variable that can be controlled MUST be
sensed. If its value is not determinable by the thing "controlling" it, then
it is not being controlled by that thing. I think you must mean something
different. Anyway, I start by assuming that the levels of intrinsic variables
can be, and must be, sensed in some way (for example, one is taught in high
school that an increase of around 5 degC usually doubles the rate of a chemical
reaction. That reaction serves as a temperature sensor).

In the initial state, there are only intrinsic variables. Nothing else is
sensed. If the entity is to survive and propagate, it acts in such a way as
to keep the levels of these variables near some genetically set reference,
which is to say that some control system is operating. Perhaps there is a
wiggle-motor driven by the aummed absolute deviation of the intrinsic variables
from genetically set reference levels. Quick wiggling gets it away from places
that drive its intrinsic variables to "bad" levels. Such a system sounds to me
like the foundation of your reorganizing system. It also sounds to me like the
foundation of the normal hierarchy.

Consider. Suppose some aspect of the intrinsic variable set tended to be
correlated with some aspect of the environment. Say, for example, that
one of the chemical chains started with the absorption of photons. It would
be advantageous if the organism had some way of detecting when it was being
bathed with photons (but not too many). Let's say that the photon-based
reaction affected the concentration of CO2 in the organism. Then if the
wiggle-motor became more specifically sensitive to deviations in the CO2 level,
both above and below optimum, then the organism would begin to control for
light level. It would, operationally, have a percept of light--part of the
normal hierarchy. But, it still would have only one degree of freedom for
control--wiggle rate--and therefore there might be conflict between setting
optimum CO2 and setting optimum values for the true intrinsic variables, for
which CO2 level was a surrogate discovered by chance. Generally (we have
assumed) good CO2 levels covary with good levels of the really important
variables, but not always. Good light levels may be found in an acid bath
lethal to the organism.

The organism would survive better if it discovered another degree of freedom
for control--say, closing and opening membrane pores, assuming it to be bounded
by a membrane. Every dissipative structure must have a way of ingesting
material and energy, and of disposing of waste, so the organism has to have the
equivalent of pores. Chemical evolution has, by now, discovered many specific
degrees of freedom for controlling such channels, so it's not an unreasonable
thing for our hypothetical wiggler-cell to discover. Given this other degree
of freedom, a reorganization driven (by your mechanism or mine) would probably
be most successful if it reduced the linkage between the intrinsic variables
and wiggle-rate, increased the linkage between light-level(CO2) and wiggle-rate,
and increased the linkage between the intrinsic variables and pore aperture.
So now we have two control systems operating in parallel, one being largely
what we might call sensory-motor (S-R terminology, forgive me), and the other
being "intrinsic," dealing only with the chemical state.

In this scenario, the reorganization affects whatever input-output connections
may be, implicit in the chemical reactions. There aren't any neural links yet.
There aren't any neurons. But there are ECSs, firstly one and then two. The
intrinsic variable levels are affected by the environmental surroundings of the
wiggler cell, such as acidity, temperature, "food," and so forth, of which light
level is one. With pore control, the wiggler cell can change its sensitivity
to these things while seeking aptimum light, but so long as the intrinsic
variable errors are not totally decoupled from the wiggle-rate, it will still
remove itself from well-lit lethal areas.

At this point, I suspect we already have a difference of opinion. If I read
your reorganization concept aright, the intrinsic error would allow the
coupling of CO2 sensing to wiggle rate, but would require something else
to be a sensory surrogate that affected the output leading to pore aperture.
And neither of these two control systems would have a reference level. I
can't see how that could work, so I imagine I do not read you aright. But
I can't see how else to interpret your decoupled reorganization and sensory-
motor control structures. And you have said that the top-level reference are
a mystery, or are always set to zero. Neither seems appropriate here.

You questioned >"How do you open up the connections from a higher to a lower

system to insert a complete control system with all its connections to and
from both the higher and the lower systems? This idea seems to me to entail
enormous difficulties, whereas building from the bottom up eliminates those
particular problems completely."

I have here proposed an example, as low-level and basic as I can imagine
(at present). In general, I assume a complex hierarchy is normally built
much as you describe, from the lowest levels of control upward, with one
important difference: always new higher levels are inserted between the
controllers for intrinsic variables and the levels already built. I assume
that it is possible but rare that an established lower level gets serously
disturbed when new ECSs are built. In your system, as I understand you, this
decoupling happens because of some (to me mysterious) organizing principle:

I'm assuming now that reorganization is specific to each level.

I think it simpler to make no such assumption. But one assumption I do make
is that each intrinsic variables has some genetically determined optimum value,
deviation from which serves as an error signal that (after amplification)
serves as a reference signal for some set of ECSs in the "normal" hierarchy.

The second assumption that I think you don't like is that each ECS, including
those involving the perception of intrinsic variables, will "reorganize" at
a rate depending on its error (or rate of change of error--the criterion is
not at issue). Reorganization could mean any of the things you talk about
as being subject to reorganization, so that's not an issue either. I am
assuming that what you don't like is the complication of the ECS by providing
a mechanism for it to reorganize. But is this more complex than having a
separate reorganization system that can go into individual ECSs like a mad
electrician and do the same rewiring job?

According to my scheme, ECSs that are able to control their percepts near their
references will not reorganize much, if at all (we agree that there may be a
threshold--I think there must be, for thermodynamic reasons). Low-level
ECSs will normally maintain control, barring accidental damage. The most
reorganization will tend to occur on higher levels that are newly being built.
Reorganization will thus SEEM to be specific to each level, but that falls
out naturally rather than being a design criterion for the system.

Occasionally, the reorganization of a high-level ECS will send to lower ones
reference levels that cannot be satisfied, causing the lower ones to experience
sustained error and to reorganize themselves. This can cascade, but is
unlikely to do so for very long in a developed hierarchy.

I think my scheme is much simpler than yours, and should be expected to occur
naturally right back to the very beginning of life/control. It is inherent in
the structure of control, rather than being dependent on two interacting
structures, one of which exists only to disturb the other. And it fulfils your
criterion that the reorganization should be random, knowing nothing about where
error comes from or what it "means" to the hierarchy.

···

------------------
Reorganizing high-gain systems:

If we think of reorganization as instituting small changes in the
parameters of control, there's no reason why a high-gain control system
can't go right on controlling. Even in category control (to pick up that
thread), slight changes in perceptual parameters would just move the
boundaries of the category a little. I think those boundaries are fuzzy
anyway, so we don't really have any discrete-variable problems here.

Aren't you shifting ground here? Hebbian perceptual learning is a very
different kettle of fish from switching the sign of an output-reference link.
The latter can't be smooth. Changing what an ECS is trying to perceive is
a little different from changing its method of trying to perceive a given
variable. Hebbian learning is important, I think, in developing control
structures that control complex environmental variables that "really" do
something coherent. If the perceptual functions are all of the simplest type,
a weighted sum followed by a nonlinearity, then the perceptual side of the
hierarchy is a multilayer perceptron, and if it has three or more layers, it
is capable of subdividing the sensory input space in any way at all. It will
divide it in a way that is consisten with coherences between the input and
the desired output.

And I don't think category boundaries can ordianrily be fuzzy. I think they
are usually catastrophic, so that although small changes usually have no
effect, sometimes they have a dramatic effect.

Look at it the other way around. If reorganization is to work as I propose,
it CAN'T work fast.

I agree, and I see that as one of its problems. My scheme can handle (will
handle automatically) the control-reversal experiment you did with Rick.
How do you explain the switch in the sign of control that happened within
half a second of the reversal of the external connection? How did the
change of control to a new, previously learned, system happen, and when and
why did that learning to reverse take place? The world isn't full of reversing
feedback connections.

I do think rapid reorganization can happen, at least within my local scheme.
Any time a ECS has a large and rising absolute error, it should be expected
to do some random sign flipping and/or reduce its gain. It is true, as you
say, that each control system sees only one degree of freedom, so if it can
flip and detect the results of the flip quickly, it can succeed through the
e-coli procedure. But, and here's a big but, its changed requirements on
lower-level ECSs may induce conflict with other same-level ECSs, causing
more error in them and possibly failing to solve its own problem. They also
may reorganize, and we are back in the same old dimensionality problem.
The error threshold for reorganization comes into play here, with luck
reducing the dimensionality to a feasible range.

----------

However we end up talking about reorganization, we have to arrange for
somatic states of the organism to have a very powerful directing effect on
what control systems are acquired. Logic and principles often bow to
hunger. How does that work? I can see how my model would do it, but how
does yours?

Hunger is presumably a surrogate for some error in intrinsic variables (I
believe it is largely caused by stomach contractions, so it isn't a sensation
of intrinsic variable state). It should be capable of control by changing
reference levels for principles or anything else. I don't know where the
actual ECSs that accept the hunger sensation as part of their input might
be in the hierarchy, but in my scheme, if hunger is not controlled over some
long period, then intrinsic variables will depart from their reference levels,
and they are the top-level ECSs, dominating all else. No problem.

Perhaps what we need to do, as we're not likely to resolve this conflict
experimentally, is to find a way to talk about reorganization so it doesn't
matter which is the right model.

There's enough in common that we should be able to do that. Reorganization
is blind and random in both cases. The only substantive difference is in the
localization of the reorganization in my scheme as compared to its
distribution in yours. That difference may often be unimportant.

Martin

[From Bill Powers (921016.0930)]

Oded Maler (921016) --

I like it. We develop a verbal means of protecting ourselves from
criticism or resistance in bumping our way among people, and use it
automatically when we bump into anything, or cause any problem. I've
sometimes thought that "pardon me" is used as another way of saying
"get out of my way."

What kind of experiment could you do to test your proposal that the
"sorry" loop is lower in the hierarchy than the conscious one?

Any more proposals out there? How about the S-R explanation?

···

-------------------------------------------------------------------
Greg Williams (921016) --

... the criteria for starting reorganization due to errors in
(some? all?) innate critical variables are subject to alteration by
reorganization triggered by sufficient error in acquired (within a
single lifetime) reference levels -- which can be thought of as
acquired critical variables.

Do I understand this correctly? You appear to be saying that a learned
system can alter not just the states of the organism that I call
critical variables (through indirect effects of its actions), but the
TARGET VALUES (criteria) toward which those critical variables are
controlled and that define zero critical error. So unless you're
junking my proposed reorganizing system, you're adding lines in the
diagram that come from the output of a learned reorganizing system and
go to the reference inputs of my proposed system. Is that what you
mean?

I know you're trying to get the house finished and do all the other
things that would keep three ordinary people busy. When things settle
down, perhaps you can get more specific about the model you're
proposing. I don't really want to go on with this until there's a
model to test. What you consider to be data depends on the model
you're using.
-------------------------------------------------------------------
Best,

Bill P.

[ Oded Maler (921023)

* [From Bill Powers (921016.0930)]

···

*
* What kind of experiment could you do to test your proposal that the
* "sorry" loop is lower in the hierarchy than the conscious one?
*

I don't have for the specific one, but in general you can let someone
play with the computer such that he/she has to respond in a certain
way to some stimulus (sic.) appearing on the screen (e.g., track it
with the cursor/mouse) within some time interval. Then make the
stimulus more complicated (a larger bit map) and change the rule of
the game such that those bitmaps divide into two categories A and B
such that for A you have to respond as before but for B you shouldn't.
But you loose much more points if you ignore A than if you
"over-react" and respond to B. Now you train the subject and compare
how the performance changes with the complexity of the categorization
(more complex distinction, e.g., degree of darkness vs. the parity of
the number of bits) together with the length of the time interval
within which he must respond. Finally you introduce an additional
activity in parallel (track completely different objects of the other
side of the screen) and see how the intensity of the latter activity
influences the over-reaction in the previous one.

But I'm not an experimentalist..

--Oded

--

Oded Maler, LGI-IMAG (Campus), B.P. 53x, 38041 Grenoble, France
Phone: 76635846 Fax: 76446675 e-mail: maler@vercors.imag.fr