Erling's papers in Boston

CSGnet Archive CSG2000
1

[From Erling Jorgensen (00.08.02.0100 CDT)]

Dick Robertson,000801.1715CDT

With allusions to:

Bill Powers (2000.06.26.1828) [his original date omitted the month]

Hello, Dick. Good to talk with you again. Thank you for your comments
and query about my papers. I hope it's okay that I post my reply to
CSGnet, instead of directly. I'm imagining there may be others
reflecting about these things. I was pretty amazed when Rick told
me Boston U. couldn't come up with a speaker phone, so we might have
discussed some of these matters more directly as a group. I'm sorry
that didn't work out, but maybe we can use the net forum to fill in
some of the gap.

I hope you got some of
your fellow students thinking about the problems of behaviorism.

Well, yes and no. I know some of my fellow counseling psych. students
were catching some of it and appreciating the points I was making about
cybernetics. The psych. prof. also seemed to understand what was at
stake with a lot of my contributions. OTOH, I circulated two e-mail
position papers on "Deconstructing a bar-press," trying to demonstrate
that it is a stability of behavioral _effect_, not of action itself, that is
produced on repeated trials. And at least the two clinical psych. students
that responded to my post couldn't see what the fuss was about. Too
far outside their current paradigm, I guess.

On pg. six of the Cybernetic Paradigm
paper, 1st paragraph, where you tackled that extremely tough problem of
consciousness...

I agree, this is a much thornier problem. As I see it, PCT speculations
about consciousness need to address themselves to such questions
as these -- (the "it" in this list is consciousness):

*When* does it arise?
*How* does it arise?
What does it *look* like when it arises?
Can it *do* anything once it has arisen?
What is it *not* doing?
What *can* it do?
Is it only *accompanying* something else which is doing the doing?
Is there an *advantage* to whatever is done?

--where you said, "I am suggesting about consciousness
[that it is] essentially an error-monitoring system...." are you saying
you think we can _perceive_ error signals?

I think I am, although there is a bit of a "Cartesian theatre" in how
you phrase your question. Do we "perceive" perceptual signals?
No, that would mean we notice rates of neural firing. Rather, our
experience of the world _is_ those perceptual signals. Afferent
signals, created by perceptual input functions, _constitute_ our
perception. This is standard PCT, articulated well by Bill [but with
an interesting caveat] in (2000.06.26.1828) -- i.e., his 2nd "hair" Re:
Outcomes and Goals.
  >> we can experience only what is represented in the form
  >> of a perceptual signal -- in general, an afferent signal,
  >> although at higher levels defining afferent can be difficult.

Bill goes on to say:

Passive awareness of perceptual signals constitutes
_consciousness_, according to the auxiliary proposals
of HPCT.

It seems to me that *passive* "experience" [calling it awareness
begs the question] of perceptual signals constitutes uncorrected
(gain of zero?) or errorless perception. And for that reason I'm
tempted to say it would be "automatic" and unconscious perception.
By contrast, the hypothesis that I am trying on for size is that
awareness uniquely arises when error comes into the picture in
some way.

It seems you might be agreeing with this, Dick. In fact, I think
I've noticed a loose consensus on the net around this (limited)
point. In your words:

I, too, suspect that consciousness is
somehow related to perceptions one is making in the systems most
affected by the largest error signals at a given moment.

This is one emerging answer to the first question in my list above.
(I.e., When does it arise?)

Back to the error signals themselves. Yes, I think I am suggesting
that consciousness _is_ the experience of error signals. [As a side
point, there could be these threshold or magnitude effects -- i.e.,
"persisting" error signals or "the greatest" error signals are what
appear in consciousness.] But you're right in your reservations.
Such would be a substantial insertion into quite a parsimonious
system of thought, so we have to be careful that it doesn't create
more problems than it might solve.

That's why I listed those eight questions above. So far, I am only
talking about the first three: When, maybe How, and What does it
look like? I am definitely *not* proposing that such (duplicate?)
error signals are routed into higher level control systems -- to create
the "parasitic feedback loops" Bill warned against when he tried to
implement a working model with such a wiring scheme.

Nor am I proposing (contra the spirit of William James' notions)
that these "conscious" error signals direct behavioral traffic by sending
reference signals directly to lower levels. Or rather, I do not mean
that there is anything about them being conscious that "sets" or
"chooses" reference signals for the lower levels. As error signals,
yes, they would channel through an output function to become a
reference for the next lower level. And if the "conscious" part is via
a reentrant loop (back into the comparator), such would constitute
amplifying the gain on that loop. Notice, the gain would go up on a
control loop that is not controlling very well, thus leading to even
greater error. [I'd have to test this on a dummy model to see whether
"greater" error or "quicker" error of "oscillating" error or something
else would be the precise effect.] But some kind of "worse" error
could play a role in localizing where the "quicker tumbles" of
reorganization should occur -- which I think might be the true role
consciousness might play in an adaptive sense.

This addresses a problem noted by both Bill P. and Bruce G. in
recent discussions on CSGnet about volition.
Bill (2000.06.30.0800 MDT) put it this way:

the E. coli type of reorganization is unspecific, and we have
always had a problem with understanding how it could be
applied just to the part of the brain that needs it without
disrupting parts that are working fine. My "reorganization
follows awareness" proposal was one attempt to increase
the specificity.

Bruce (Fri, 30 Jun 2000 12:40:05) remarked in a similar vein:

It seems unlikely to me that the system to be altered is a
random choice. It is probably the system with the greatest
error and systems subordinate to it. The direction of change
is likely to be [random].

So then, what I am proposing about consciousness is one way to
address and further refine the notion of directed reorganization.
What is tried (in terms of altered parameters) remains random,
but *where* it is tried is focused on systems with large or increasing
error. So, yes, consciousness can indirectly "do" something, and
what is done does have potential adaptive advantage for the
organism, in terms of speedier reorganization where it is needed.
It is ironic to consider, under this proposal, that consciousness
makes it worse so it can get better sooner! Maybe that holds
some parallels to how you have thought about anxiety and the
Method of Levels.

Let's return to the third question in my list, because I think one of
your other comments may be referring to that -- i.e., What does it
look like when it arises? You say:

But in current
PCT thought at least you have to have a sense receptor in order to
perceive anything. And that is where I have hung up on the idea. I
can't imagine what a sense receptor for an error signal would be like.

I'm trying to think like a strict PCTer here, so my apologies if this is
just being picky. I'm not quite sure what you mean by "sense receptor"
here. Is it the perceptual input function *creating* the perception, or
the perceptual signal channel *carrying* the perception? Is there
anything more than these two that you would need, in order to perceive?
I don't want any stray homunculi around, needing to "decipher" or
"reconstruct" a given perception. Once you've built it with a perceptual
input function, you've got it available for whatever you might need.

And it seems to me that to subtract a reference signal from such built
up perceptions (or vice versa) does not necessarily alter the "qualia"
of those signals. After all (IMHO), the comparator doesn't label or
differentiate a reference from a perceptual signal; it just outputs the
difference, so it can have further effects downstream. Now, you are
right that at _some_ point in the loop, (according to standard elegant
PCT), we no longer perceive, we just do something with the resultant
signal. For instance, we do not seem to perceive motor outputs, only
their perceptual effects.

And you could well be right, that what I'm proposing needs more than
a simple subtractive comparator as an "error input function" for errors
to be "experienced" in some conscious sense. (Or at least maybe it would.)
I suppose one way to reconcile our respective thoughts on this would
be to consider whether mounting error could have any "potentiating"
effect on the (nearby) inputting perception, such that it "reverberates"
or stands out consciously from the background of all other perceptions.
[This would be an alternate answer to the How does it arise? question.]

A question we haven't yet posed is -- How does it *stop* being
conscious? Either "potentiated/reverberating perceptions" or
"reentrant/amplifying error signals" raise the problem of why those
specific systems do not _persist_ in consciousness, after the
increasing errors have been reorganized away. I do not think
it is sufficient to consider that some other error-increasing system
surfaces into awareness, ("drowning out" other signals clamoring
for consciousness?), because under any of these speculations,
the previous consciousness-carrying channels are still there.
Perhaps it is enough to consider that anything we have at one
time become conscious of, we can again ponder consciously,
or that it persists in a low-grade (almost subliminal) conscious
mode.

One further point. These notions do not address Bill's ideas
about volition, as a way of injecting arbitrary reference signals
into the hierarchy just to "see what would happen." While I
like the problem-solving and conflict-resolving potential of
such a notion, I can not yet imagine a mechanism capable of
doing that. Nor do I think what I am suggesting about
consciousness provides such a mechanism. So a tighter
integration of notions about volition and consciousness
remains for more fruitful minds than mine to consider.

Thanks for you interest and reflections. All the best.
        Erling

2

[From Dick Robertson,000802.1530CDT]

Erling O Jorgensen wrote:

[From Erling Jorgensen (00.08.02.0100 CDT)]

> Dick Robertson,000801.1715CDT
With allusions to:
>> Bill Powers (2000.06.26.1828) [his original date omitted the month]

Hello, Dick. Good to talk with you again. Thank you for your comments
and query about my papers. I hope it's okay that I post my reply to
CSGnet, instead of directly. I'm imagining there may be others
reflecting about these things.

I'm sure.

> I hope you got some of
> your fellow students thinking about the problems of behaviorism.

Well, yes and no. I know some of my fellow counseling psych. students
were catching some of it and appreciating the points I was making about
cybernetics. The psych. prof. also seemed to understand what was at
stake with a lot of my contributions. OTOH, I circulated two e-mail
position papers on "Deconstructing a bar-press," trying to demonstrate
that it is a stability of behavioral _effect_, not of action itself, that is
produced on repeated trials. And at least the two clinical psych. students
that responded to my post couldn't see what the fuss was about. Too
far outside their current paradigm, I guess.

Welcome to the club.

> On pg. six of the Cybernetic Paradigm
> paper, 1st paragraph, where you tackled that extremely tough problem of
> consciousness...

> --where you said, "I am suggesting about consciousness
> [that it is] essentially an error-monitoring system...." are you saying
> you think we can _perceive_ error signals?

I think I am, although there is a bit of a "Cartesian theatre" in how
you phrase your question. Do we "perceive" perceptual signals?
No, that would mean we notice rates of neural firing. Rather, our
experience of the world _is_ those perceptual signals. Afferent
signals, created by perceptual input functions, _constitute_ our
perception.

Yes, but remember that in B:CP those perceptual input functions begin with a
sensory receptor. I was asking whether you were positing some kind of sensory
receptor for error signals.

Keep on slugging. It'll be worth it in the end I'm sure.

Best, Dick R.

3

[From Bill Powers (2000.08.03.0522 MDT)]

Erling Jorgensen (00.08.02.0100 CDT)--

... Do we "perceive" perceptual signals?
No, that would mean we notice rates of neural firing. Rather, our
experience of the world _is_ those perceptual signals. Afferent
signals, created by perceptual input functions, _constitute_ our
perception.

That may be true, but it does not follow that when we are _not_ consciously
perceiving (experiencing) X, there are no neural signals representing the
state of X. The implication that experience is identical to the existence
of neural signals is, I claim, incorrect. Perceptual signals can exist
without our experiencing them.

This is most easily seen when our attention is called to a perception that
_we must have been controlling before we consciously noticed it_. This is
most easily done by directing attention down or up a level. Example: you're
sitting on the floor in an awkward position, trying to get the tip of a
screwdriver into the slot of a screw you can barely see on the underside of
a kitchen cabinet. I point out to you that your unused hand is pushing
against the floor to hold your body in place. Chances are that you were
not, at that moment, consciously perceiving that fact, but began
experiencing it as soon as I mentioned it. However, since you were holding
your body in place by means of applying this pressure to the floor using a
known control system, the perception must have existed before you noticed
it, because perception is necessary for control.

Many examples occur during MOL sessions: a person will describe a
higher-level perception without consciously noticing doing so; the guide
calls attention to this, and at that moment the perception _becomes_ an
experience in consciousness.

This type of phenomenon seem to suggest a difference between perception
(existence of signals in a perceptual channel) and awareness (consciously
experiencing these signals). It's as though something is receiving (or not
receiving) copies of the perceptual signals. When there is no perceptual
signal there is, of course, no experience. But presence of a perceptual
signal does not automatically create an experience. Something more than
presence of the signal is necessary.

By contrast, the hypothesis that I am trying on for size is that
awareness uniquely arises when error comes into the picture in
some way.

That is a possibility, but (as your careful wording suggests) it does not
necessarily imply that awareness is _of_ the error signal. What happens
when an error signal suddenly appears in a control system somewhere in the
hierarchy? The immediate effect (in a few milliseconds) is that subsidiary
reference signals change. As a result, your actions change, and as a result
of _that_, you _perceive_ your actions changing. When your foot starts to
slip on an icy sidewalk, the first you know of it is the violent moves made
by your other leg and your arms as your lower control system are used to
maintain your balance. Those moves are directly experienced evidence of
error signals, but they are not error signals. They are perceptions. You're
witnessing the effects of your own control systems in action, but you're
not consciously deciding to operate them. They operate by themselves. By
the time you could consciously evaluate the situation, you would either be
balanced again, or flat on your back.

Consider the Case of the Crooked Picture. You're walking down a hall and
you notice that a picture on the wall is tilted. If you have no active
reference condition relating to how pictures hang, there will be no error
signal. You may note that the picture's frame makes a small angle with the
sides and corners in the hall, but this will not call for any action; you
will walk on. But if you do demand that pictures always hang straight,
there will be an immediate error signal. Instantly, at the speed of
propagation of neural signals, the error signal will be converted into
reference signals that ordinarily would stop your motion, raise your hands
to grasp the frame, and tilt the picture back into parallelism with the
hall -- if you had no other goals at that moment. More likely, you would
feel a little urge to straighten the picture but other concerns of more
importance would dominate and you would just keep going.

You might _think_ you had experienced an error signal in all these
circumstances, if you knew PCT, but how could you really tell the
difference between the error signal and all the immediate lower-level
effects it creates? We can easily explain how changes in reference signals
could immediately lead to changes of perceptions, but what basis could we
offer for proposing that in addition to the perceptual signals, we would
also experience the error signals?

One of the common mistakes made in the past was to confuse the target in a
tracking task with the reference condition. In fact it's easy to show that
the perceived distance between the target and the cursor is NOT the error
signal, just by asking the participant to choose some reference condition
other than zero distance. When this is done, and it's easy to do, the
cursor remains some fixed distance from the target as the disturbances
vary. Zero error now corresponds to a nonzero distance. When you actually
try this, it becomes obvious that there is nothing in experience that
corresponds to the error signal. You have to imagine the error signal,
knowing that it's the difference between the desired condition and the
actually-perceived condition. If you didn't know about PCT, why would you
imagine an error signal? Note, by the way, that you could walk down a hall
systematically tilting all the pictures to make a specific angle with the
walls and corners.

Your eight questions about consciousness all seem to imply that it's
something outside us that we can examine. I don't think it is. We can't see
anyone else's consciousness, and when it comes to our own, all we can do is
experience those perceptions we experience. Evidence about consciousness
(that is, observations we can put into some deductive scheme) is hard to
find: we can't really get outside ourselves to watch ourselves being
conscious. The most direct evidence is the way certain elements of
experience can suddenly appear as if from nowhere, although we know
rationally that they must have been present before they appeared. When I
point out the pressure of the seat you're sitting in, you probably don't
believe that you acquired a bottom just at the instant I mentioned it. It's
harder to notice when things fade out of immediate experience, but it's
hard to ignore the phenomenon when they return. Trying to observe the
Observer is fruitless, but with the aid of memory we can observe _having
observed_ just a moment ago, and that is almost as useful.

I have spoken with people who seem not to know that they are observing. If
I were to believe what their words appear to say, it would seem that for
them there is simply a world of experience without any need for an observer
to note its presence. In fact, they speak as if the world they experience
is the actual objective world as it really is. They tend, of course, to be
Gibsonians, because unless you can explain how it is that we know about the
physical world, perception becomes a serious problem.

When you experience yourself as an Observer, the relation between observer
and observed becomes obvious. The nameless sense of being in the position
of the observer isn't, however, normally in the forefront; special
circumstances are needed to bring out the phenomena of awareness and
consciousness. No special circumstances are needed to _create_ the
phenomena, but only to create the kind of evidence that our higher-order
systems need in order to think about them. We have to teach oiur own mental
models to take the Observer into account, so our rational processes do not
assume they are the highest ones in the brain.

Best,

Bill P.

4

[From Bruce Gregory (2000.0803.1227)]

Bill Powers (2000.08.03.0522 MDT)

That is a possibility, but (as your careful wording suggests) it does not
necessarily imply that awareness is _of_ the error signal.

As I understand the HPCT model, the reorganization system must have some way of
detecting the presence of persisting error. In that sense, at least, there must
be a signal that invokes reorganization. Or, if reorganization is ongoing, that
directs reorganization to the subsystem that is encountering persisting error.
But, as you say, this does not have to involve awareness of the error signal.

When you experience yourself as an Observer, the relation between observer
and observed becomes obvious. The nameless sense of being in the position
of the observer isn't, however, normally in the forefront; special
circumstances are needed to bring out the phenomena of awareness and
consciousness. No special circumstances are needed to _create_ the
phenomena, but only to create the kind of evidence that our higher-order
systems need in order to think about them. We have to teach oiur own mental
models to take the Observer into account, so our rational processes do not
assume they are the highest ones in the brain.

The self you experience as Observer seems to me to have no other "properties".
Memory, ability, and will seem to reside in the hierarchy. In that sense, we (as
Observer) cannot teach our models anything. It seems to me that what must occur
is that the hierarchy be able to solve some problem by adopting the belief that
"you" are the observer. In this case, the hierarchy will adopt this belief.
Otherwise it will simply dismiss the argument--or deny that it experiences the
Observer at all!

BG

5

[From Erling Jorgensen (00.08.03.1400 CDT)]

Dick Robertson,000802.1530CDT

Erling Jorgensen (00.08.02.0100 CDT)
our experience of the world _is_ those perceptual signals. Afferent
signals, created by perceptual input functions, _constitute_ our
perception.

Yes, but remember that in B:CP those perceptual input functions begin with a
sensory receptor. I was asking whether you were positing some kind of
sensory receptor for error signals.

Not at this point, I'm not. Such a change could introduce too many of those
unforeseen ramifications into the model, that Bill was talking about. I'm
trying to stay very close to what is already in the model. Small changes
can have large effects, in particular, unknown ones.

We already have an "error creation function" -- it's called the comparator.
What we don't have is a "location" for conscious experience. I was trying to
see whether a copy of the error signal could (somehow) provide a substrate
for consciousness. [It's a simplistic proposal, I know, but the HPCT models
have been amazing in showing how much can be achieved with a very
bare-bones approach.] In addition, there are phenomenological intuitions
which many (including yourself) have mentioned, that consciousness seems
to have some relation to systems encountering sufficient or sustained error.

The first problem that arises with letting a copy of an error signal "be
conscious" is where does that copy go and what does it do? And does it
create more problems than it solves, when you let it go there and do that?
Bill has warned of severe modelling problems that arise if you let that
error copy go to higher levels. And we already have nicely operating
lower levels that make use of the original error signal, which we don't
want to foul up by sending errant error signals here and there.

A safer possibility is having the conscious error signal reenter the
comparator. We already know that the output of the loop has to have
its gain amplified in some way, so this would seem to introduce no new
properties into the model. [Note, however, that _all_ loops have high
amplification in gain, but not all loops are necessarily conscious,
(it seems the vast majority are not in their normal functioning),
so this should not be the only way to amplify gain.]

But notice one of the payoffs of having a poorly functioning loop (i.e.,
one with high error) amplified: its control and its error becomes worse,
and this could be a way of "signalling" reorganization where its services
are needed. As yet, in the model, we have had no explicit mechanism for
directing reorganization to those systems that need it most. But several
people have noted phenomenologically that while what we try may be random,
where we try it generally is not. So with a simple proposal of reentrant
error signals, we may have stumbled upon a way to tie up a loose end in
our understanding of reorganization. The effect of those reentrant paths
when error is low would simply be to increase the sensitivity of those
loops in responding to whatever error is present; so again, no new
properties have been introduced.

This does not yet speak to whether we should consider such reentrant paths
"conscious" or not. Martin has argued that awareness is awareness-of-error,
and I probably lean in that direction. Becoming aware of something (even by
way of a verbal suggestion to attend to a previously unconscious body
configuration) seems to entail "error-introduction" -- either a relatively
sudden disturbance from the environment (e.g., bumping an elbow) or a sudden
increase in the reference for perceiving that configuration (e.g., "What
angle is your elbow at?). It remains to be seen whether for any given
instance of awareness an identifiable error component could be found.

I had been reasoning sort of by analogy -- (not the best way to proceed)
-- that if a signal in a afferent channel could be perceived, a modified
version of that signal just dowstream (i.e., an error signal) could be
"aware." However, you raise a good point with your mention of sensory
receptors. There may well be a qualitative difference that occurs just
at that spot in the loop. Bear with me, as I try to reason this out.

When it comes to the sensory receptor, we are not just noting the same
neural energy which has been circulating in the loop, just manipulated
in a new way. We are dealing with a transformation of energy from one
kind into another. At all times, there are various energy gradients
impacting us from the environment. Sensation in a generic sense involves
first of all a transformation of that energy -- e.g., from sound waves
into mechanical action (I think?) and then into neural firing in the ear,
or from light into neural firing in the eye -- and that is what you raise
with the question about sensory receptors. That is the first "difference
that makes a difference," to use Bateson's phrase.

After that, there are further differences that seem to make a difference.
For instance, there are recombinations and weightings of those first
neural energies into subsequent perceptual patterns, which allow new layers
of control to be built up. But you may be right in your implication that
neural energies by themselves are not sufficient for "perception" to occur.
After all, the motor output pathways also use neural energy, in various
weightings and recombinations, and we do not seem to "perceive" those signals.

I had been thinking that what makes perception perception, is *where* it
occurred -- in afferent pathways, (or perhaps in certain areas of the brain)
-- or maybe in *how* it was constructed by the various perceptual input
functions. But a more crucial distinction may be that transformation from
one kind of energy into another, that occurs at the point of those sensory
receptors. What we perceive are the changes (and invariants?) in energy
from the environment, *as they are transformed* by the sensory receptors.
Even though there are many subsequent modifications of the neural energy
that results, the perceptions that are thereby constructed do *not* seem to
be "in here" where the perceptual input functions are operating, but "out
there" in the environment where the initial energy changes are occurring.

In other words, the difference that makes a difference for perception [as
I am now thinking about it] is the *transformation* of energy that occurs
at the interface of the environment and the organism. After that, the
perceptual input functions create a multiplicity of virtual perceived
environments, with which each next higher level interacts. Those functions
entail various *modifications* of the neural energy into new patterns, and
they "make a difference" for new types of control. But their "qualia" as
perception has to do with the feeding and transformation of energy from
the external environment; [or so I am now wondering].

I am not sure how this affects your original question. I am extrememly
reluctant to "posit some kind of sensory receptor for error signals," even
though the view of perception in these last few paragraphs leans the analysis
that way for them to be "experienced as conscious." This distinction between
"transformed energy" and "modified signals" is a little too flimsy to insist
that we should start looking for energy transformation occurring in and
around the comparator. Components such as electrical potentials and chemical
neurotransmitters might provide substrates for such a distinction, but I'd
hate to go speculating about all this at that level of specificity.

I think it would help me if someone could lay out the "wiring" for
passive-but-aware perception (one of the alternate proposals for
consciousness). Does this mean:

        Reference-of-zero?
        Gain-of-zero?
        A-perception-not-yet-corrected?
        Is this G. Gordon Liddy's trick ("The-secret-is-not-to-care")?
        Is this the Zen way (aware-of-the-flow-but-uninvested)?
        Is it "the Observer" (with a potential Dead-Sea-reservior of any
perceptual signal but no outlet channel)?

I think I'm trying to get at here other people's views of "What does it
look like when it does arise?" and does that overlay anything on our
current model. For myself, I still like the notion that consciousness
is somehow connected to sudden or sustained error. And from my counseling
background I retain a bias -- as I mentioned in my last post -- that
"consciousness makes it worse so it can get better sooner."

Keep on slugging. It'll be worth it in the end I'm sure.

I am, and I'm enjoying the workout. Some of the slugging, however, was in
the rest of that previous post. Any reactions as yet to those portions?

All the best,
        Erling

6

[From Erling Jorgensen (2000.08.04.0900 CDT)]

Bill Powers (2000.08.03.0522 MDT)

Erling Jorgensen (00.08.02.0100 CDT)--

... Do we "perceive" perceptual signals?
No, that would mean we notice rates of neural firing. Rather, our
experience of the world _is_ those perceptual signals. Afferent
signals, created by perceptual input functions, _constitute_ our
perception.

That may be true, but it does not follow that when we are _not_
consciously perceiving (experiencing) X, there are no neural signals
representing the state of X. The implication that experience is identical
to the existence of neural signals is, I claim, incorrect. Perceptual
signals can exist without our experiencing them.

I fully agree. I didn't think anything in either of my recent posts implied
that I think otherwise. (I think my use of "experiencing" is more generic,
not restricted to conscious experience, if that's where the confusion if
arising.)

This type of phenomenon seem to suggest a difference between perception
(existence of signals in a perceptual channel) and awareness (consciously
experiencing these signals). It's as though something is receiving (or not
receiving) copies of the perceptual signals.

Yes. In my second post (00.08.03.1400 CDT), I tried to address the issue
with copies -- where do they go and what do they do when they get there?
The answer I explored was reentrant paths. To do that with a perceptual
signal prior to the comparator would seem to skew too many things in the
functioning of the loop. So I'm exploring doing that with the error signal
(functionally equivalent to amplification of gain, so no new properties.)
There seem to be benefits of considering reentrant error signals (e.g.,
"signalling" reorganization), even if we do not experience such signals
consciously.

By contrast, the hypothesis that I am trying on for size is that
awareness uniquely arises when error comes into the picture in
some way.

That is a possibility, but (as your careful wording suggests) it does not
necessarily imply that awareness is _of_ the error signal.

I am aware of that (sorry for the pun ;-). Yes, these parts of the proposal
are distinct -- awareness can accompany error without being _of_ the
error. We can (provisionally) accept or reject one or both parts.

What happens
when an error signal suddenly appears in a control system somewhere in the
hierarchy? The immediate effect (in a few milliseconds) is that subsidiary
reference signals change. As a result, your actions change, and as a result
of _that_, you _perceive_ your actions changing.

Yes, consciousness lags behind control (if it appears at all). That is
true for either of our proposals -- awareness of/as reentrant error signals
(mine) or awareness of lower level effects of action (yours). Experientially,
yours is more plausible (even to me 8-o). In terms of modeliing, I still
don't know what is "done" with the copies of those perceptions.

When your foot starts to
slip on an icy sidewalk, the first you know of it is the violent moves made
by your other leg and your arms as your lower control system are used to
maintain your balance. Those moves are directly experienced evidence of
error signals, but they are not error signals. They are perceptions. You're
witnessing the effects of your own control systems in action, but you're
not consciously deciding to operate them.

Agreed. Whatever consciousness "does," it is *not* to directly operate
control systems or "decide" on anything.

As a sidebar, your thoughts about volition -- i.e., injecting arbitrary
reference signals into the hierarchy just to see what would happen --
does have implications for "deciding" (or resolving an impasse), in those
situations when effects of various choices can not be anticipated. And I
think you are viewing volition as some kind of "output" of conscious
channels (??) But your point above remains: consciousness is not the
standard method of control -- it comes too late to affect the normal
process.

You're walking down a hall and
you notice that a picture on the wall is tilted. If you have no active
reference condition relating to how pictures hang, there will be no error
signal.

Is "no active reference signal" the same as a reference of zero? (I suspect
not.) To get "no error signal," wouldn't we need to have an inhibited copy
of the perception arriving at the comparator at the same time? In effect
this would make the "reference" a floating (inverted) copy of the perception
-- i.e., "Show me whatever is there."

I realize that another way to get "no action" _despite_ a non-zero error
signal is by having a gain of zero. Any systematic way of raising and
lowering gains? Does the highest gain always predominate, or is that
only in situations of direct conflict? In other situations, does it then
"come first"?

One of my ongoing references here (as you may have noted) is to understand
a taxonomy of what actions and effects go (and do not go) with what changes
in signals and parameters in the model. Obviously, as the discussion about
consciousness indicates, another reference is: "Is everything accounted for?"

how could you really tell the
difference between the error signal and all the immediate lower-level
effects it creates? We can easily explain how changes in reference signals
could immediately lead to changes of perceptions, but what basis could we
offer for proposing that in addition to the perceptual signals, we would
also experience the error signals?

Good questions. The phenomenological evidence is the weakest link, for
thinking we experience error signals. The first task, however, was to see
whether such a proposal violated any existing properties of our working
model, and to check whether it created more havoc than it promised to
resolve. (How's the scorecard doing so far?)

Your eight questions about consciousness all seem to imply that it's
something outside us that we can examine.

No, I think it's clear that much of this requires imagination mode --
it's a map of a spotty map of a map of a territory. We control, so we
must be able to perceive. That's the map of the territory all of us use
all the time. And incidentally, each of us has "our own map." We *seem*
to be able to survey *some* of those perceptions on our map, especially
important (i.e., error-ridden) ones. It leads to a spotty map. So given
that, what does it mean to be a "surveyor"?

BTW, we can exist in the territory just fine, and even do some usefully
selective surveying right where we need it, without knowing the answer
to that question.

As usual, thanks for the interchange. All the best.
        Erling

7

[From Bill Powers (2000.08.05.1416 MDT)]

Bruce Gregory (2000.0803.1227)--

The self you experience as Observer seems to me to have no other

"properties".

Memory, ability, and will seem to reside in the hierarchy. In that sense,

we (as

Observer) cannot teach our models anything. It seems to me that what must

occur

is that the hierarchy be able to solve some problem by adopting the belief

that

"you" are the observer. In this case, the hierarchy will adopt this belief.
Otherwise it will simply dismiss the argument--or deny that it experiences

the

Observer at all!

Yes, and I don't understand quite how it is that I can know anything about
the existence of an Observer. Knowing is a function of the hierarchy, as I
see it. I speak now as a logical, reasoning entity, and by my own logic I
know, or think, that there is an entity who is sometimes with me and
sometimes not. According to some theories, this entity is God. There's too
much wrong with that theory, however, to attract me. I feel encouraged to
conclude that I am part of this entity. I can't observe it but it can
observe me. I don't know why I think that.

Looking from a different platform, I see that the logical I confirms my
sense of being an experiencer, yet also being the experienced one.

Then there is the sense of having just returned from another place. And I
say, with the sense of passing on a message: I am the Observer, too. But I
am not saying that. Logically, and in principle, and in a way that suggests
a way of being, I can make some sense of all this. But not much. It doesn't
really merit the term "theory".

The method of levels is based more or less on this collection of hunches.

Best,

Bill P.

8

[From Bill Powers (2000.08.04.1450 MDT)]

Erling Jorgensen (00.08.03.1400 CDT)--

A safer possibility is having the conscious error signal reenter the
comparator. We already know that the output of the loop has to have
its gain amplified in some way, so this would seem to introduce no new
properties into the model. [Note, however, that _all_ loops have high
amplification in gain, but not all loops are necessarily conscious,
(it seems the vast majority are not in their normal functioning),
so this should not be the only way to amplify gain.]

I think I'm going to let that sit there. Modeling calls once again, and I'd
like to slip back into that mode. It won't hurt your ideas to let them hang
for a while.

Best,

Bill P.

9

[From Erling Jorgensen (2000.08.05.0940 CDT)]

Bill Powers (2000.08.04.1450 MDT)

Erling Jorgensen (00.08.03.1400 CDT)--

A safer possibility is ...(etc.)

I think I'm going to let that sit there. Modeling calls once again, and I'd
like to slip back into that mode. It won't hurt your ideas to let them hang
for a while.

That's fine with mine. You're right; it won't.

My immediate goal was to put these ideas forward, so people could
consider them. Between the papers and the last few days of e-mail
clarifications, that has been done. A future goal will be to consider
publishability -- especially of the first paper, and maybe a modified
form of the second. To that end, I would invite further comment
from anyone on the net, in particular, ideas about what venues to
consider for publishing. Thanks.

All the best,
        Erling