The culture thing

[Francisco_Arocha]

[from francisco arocha, 980319.9:03 AM EST]

Please, let me know is this make sense. I see two viewpoints in the
culture thread. One is that from the standppoint of the individual,
culture is obviously a systems concept --if the theory is right. In any
case, it is a perception.

However, culture is more than a perception when looked from a
supra-individual point of view; that of a sociologist or an
anthropologist, or whoever is interested in such things. These people are
interested in culture, not only as a perception, but also as the intended
or unintended results of human action. An example is the crowds or
gatherings programs, where, for instance, the formation of an arc or a
circle of people around a person (a street performer, a religious leader
or whatever) is the result of each individual controlling his or her own
perceptions that have nothing to do with the perceptions of arcs or
circles. These happen to be side effects of their perceptual control of
other variables.

But they also happen to be emergent phenomena, since they don't exist at
the individual level (a single individual cannot form an arc around a
"leader"). This is the same for any cultural thing or event (a burocracy
has a "rigid" hierarchical structure, which is an emergent phenomenon of
each individual controlling their own perceptions). It is these emergent
things or events, as objective happenings out there in the world, that
the sociologist if interested in.

An explanation based on PCT is not necessarily reductionst, because it
does not negate that arcs or circles exist in crowds or that hierarchical
structures exist in burocracies, as objective things, although it negates
that these are controlled perceptions. Also, PCT, as in the crowds
programs, presuposes that there is a level of reality that is
supra-individual, but one which is explained at the lower psychological
level. The question is whether these cultural things constrain, sort of
like physical reality, the perceptions that people control. But this is
an empirical issue.

francisco

[Bill_Powers1]

[From Bill Powers (980319.0916 MST)]

francisco arocha, 980319.9:03 AM EST--

Please, let me know is this make sense. I see two viewpoints in the
culture thread. One is that from the standppoint of the individual,
culture is obviously a systems concept --if the theory is right. In any
case, it is a perception.

Your whole post make sense to me. A social system, as you say, exists
because people perceive and control for variables at the appropriate
levels. But it is also emergent from interactions, as you say, and
therefore it (a) has structure, and (b) creates constraints within which
each individual must operate.

Emergence seems to be an acceptable rebuttal to accusations of
reductionism. An emergent phenomenon (like arcs and rings) is emergent from
interactions among independent control systems, but can't be explained in
terms of the properties of any one control system. As you say, one person
can't form an arc or ring around another, and the arcness or ringness is
not a property of any person.

I get the feeling that my statements to the effect that social systems are
not control systems is being taken to mean that social systems are
unimportant or have no effects on behavior. I hope this gets sorted out.

Best,

Bill P.

[bnhpct]

[From Bruce Nevin (980319.1325 EST)]

Bill Powers (980319.0916 MST)]

I get the feeling that my statements to the effect that social systems are
not control systems is being taken to mean that social systems are
unimportant or have no effects on behavior. I hope this gets sorted out.

There are at least three distinct threads commingled here.

o Social systems are not control systems.

o This or that account of things with PCT is reductionist.

o Cultural systems and language structure are not simply emergent phenomena.

Emergence seems to be an acceptable rebuttal to accusations of
reductionism. An emergent phenomenon (like arcs and rings) is emergent from
interactions among independent control systems, but can't be explained in
terms of the properties of any one control system. As you say, one person
can't form an arc or ring around another, and the arcness or ringness is
not a property of any person.

Arcs and rings are formed pretty much the same anywhere that control
systems reduce distance from a point of interest and avoid reducing
distance from their fellows below some minimum. Are they not formed by
simians? Very likely by lower orders of animals? To the extent that they
are universal, they are not cultural.

When you get people controlling a perception of how they appear to others
in comparison to how others appear to them, then you start to get cultural
phenomena.

This is the distinction I am making when I say that cultural systems and
language structure are not simply emergent phenomena, whereas rings and
arcs are.

Perhaps I am wrong in this. Are they emergent phenomena of a more complex
nature, but emergent phenomena nonetheless? Emergent phenomena are
unintended byproducts. Even when individuals participating in e.g. a ring
or arc perceive what they are doing, they are not controlling those
perceptions. This can change in group dancing, parade formations, teacher
telling children to form a circle, etc., but that is a separate matter.
Rings and arcs are multi-individual phenomena. No one individual can form a
ring.

But individuals can control their manner of doing a thing that "everyone"
does. Smiling is universal. Smiling so that the upper teeth are exposed but
the lower lip covers the edges of the upper teeth, or so that the lower
teeth are exposed but not the upper, and other variants, are norms
characteristic of different regions of the United States. (And everywhere
else of course, but that's what I've seen data for. And setting aside that
individuals of course can prefer to control perceptions other than
"reflecting the norm" if e.g. self conscious about bad teeth.)

Observe a child unconsciously affecting mannerisms of a parent, or of a
neighborhood friend. Is this an emergent phenomenon? I don't think so. What
is it? It does not call for adding any new level to the hierarchy. Perhaps
it does call for a new mechanism of control or a new aspect of control. In
PCT, we assert that behavioral outputs are not controlled, they are the
(uncontrolled) means of controlling perceptual inputs. With cultural
matters, it appears that control loop B receives as input perceptions of
the behavioral outputs of control loop A and compares them with remembered
perceptions of other people's behavioral outputs. Without interfering with
control in loop A, control loop B varies the manner in which control of A
is accomplished, and makes it resemble a remembered manner of doing so.

Is this not something new and not previously modelled? Does this not bear
perhaps on learning and reorganization?

  Bruce Nevin

[Bill_Powers1]

[From Bill Powers (980220.0323 MST)]

Bruce Nevin (980319.1325 EST)--

When you get people controlling a perception of how they appear to others
in comparison to how others appear to them, then you start to get cultural
phenomena.

This is the distinction I am making when I say that cultural systems and
language structure are not simply emergent phenomena, whereas rings and
arcs are.

I guess I don't see what alternative you're offering. If cultural systems
(composed of whole human beings and their artifacts) and language are not
simply emergent phenomena, then what _are_ they? Are you saying that there
is some kind of organizing principle operating on a nonphysical plane?

Smiling is universal. Smiling so that the upper teeth are exposed but
the lower lip covers the edges of the upper teeth, or so that the lower
teeth are exposed but not the upper, and other variants, are norms
characteristic of different regions of the United States. (And everywhere
else of course, but that's what I've seen data for. And setting aside that
individuals of course can prefer to control perceptions other than
"reflecting the norm" if e.g. self conscious about bad teeth.)

But so what? In the terms of the present discussion, we have to ask _why_
smiling is universal. How is that different from saying that walking is
universal, or digesting?

Observe a child unconsciously affecting mannerisms of a parent, or of a
neighborhood friend.

Why do you say "unconsciously"? It seems to me that children very
consciously imitate their parents and other adults, often hilariously and
often in ways that are not intended to be funny.

Is this an emergent phenomenon? I don't think so. What
is it? It does not call for adding any new level to the hierarchy. Perhaps
it does call for a new mechanism of control or a new aspect of control. In
PCT, we assert that behavioral outputs are not controlled, they are the
(uncontrolled) means of controlling perceptual inputs. With cultural
matters, it appears that control loop B receives as input perceptions of
the behavioral outputs of control loop A and compares them with remembered
perceptions of other people's behavioral outputs. Without interfering with
control in loop A, control loop B varies the manner in which control of A
is accomplished, and makes it resemble a remembered manner of doing so.

That sounds reasonable to me. But aren't you arguing on my side, that these
processes are occurring entirely within the individual? It seems to me that
you're offering a perfectly good analysis of how imitation might work,
using nothing but existing elements of HPCT.

Is this not something new and not previously modelled? Does this not bear
perhaps on learning and reorganization?

I think something like this explanation of imitation has been offered
before, but that doesn't invalidate your analysis. It doesn't require going
outside the boundaries of HPCT, as far as I can see. I'd say you have a
tenable proposition here, so now the question is how to test it to see if
we have to accept it.

I don't see why your proposition does NOT concern emergent phenomena. It
doesn't seem to call for anything that the hierarchy hasn't already been
proposed to do. Perhaps you could be more explicit about just what new
processes you're talking about.

Best,

Bill P.

[bnhpct]

[From Bruce Nevin (980320.1022 EST)]

Bill Powers (980220.0323 MST)--
(I think 980320 was intended)

That sounds reasonable to me. But aren't you arguing on my side, that these
processes are occurring entirely within the individual? It seems to me that
you're offering a perfectly good analysis of how imitation might work,
using nothing but existing elements of HPCT.

I have no interest in using anything more than existing elements of HPCT.

There are two sorts of processes. One sort occurs entirely within the
individual (within ego), control of perceptual input. That which the
individual is observing and modelling is occurring outside individuals, ego
included. Behavioral outputs are a byproduct of perceptual control and so
have no status as a separate process, you say; it is an emergent
phenomenon, of no more significance than the rings and arcs observed in the
gather program.

It does have status as a separate phenomenon because the individuals
involved are controlling perceptions of it as a phenomenon in their
environments. Individuals forming rings and arcs are not controlling
perceptions of the rings and arcs that they form. Individuals constraining
their manner of doing things in relation to observable norms are
controlling perceptions of the norms that are thereby reflected in their
behavioral outputs.

This has been called rule-related behavior as opposed to rule-governed
behavior. Departures from norms, inclusions of something from a different
system of norms (deliberately saying "ain't" for a simplistic example), and
other dances around and through systems of norms, are all ways of signifying.

You dislike descriptive generalizations because they do not apply to any
individual at any time. But what is happening here is that the individuals
are setting their own reference levels of controlled perceptions according
to their own descriptive generalizations about observed congruities of the
behavioral outputs of their fellows. The congruities emerge because their
fellows are all doing likewise. The generalizations are not without
exception and rarely reach closure (with the caveat that we have little
appreciation today of what it is like to live in a small, stable,
monocultural community).

I'm sorry if this seems to you vague or "sloppy," I'm still groping for
ways to articulate what I know about language and culture starting from
first HPCT principles. Perhaps all of this is no more than a matter of
"emergent phenomena." Processes of control are creating something new in
the environment, not accounted for by physics. Sounds like an emergent
phenomenon. What I am resisting is the proposition that it is as simple a
matter as the rings and arcs, for the reasons that I said.

I'd say you have a
tenable proposition here, so now the question is how to test it to see if
we have to accept it.

I'll keep working on that. Help from anyone would be appreciated.

Please keep this discussion separate from any speculations about
supra-individual levels of control--something that I believe is for
principalled reasons not perceivable within our perceptual universe, just
as vertebrates' perceptions are *necessarily* not available to the cells
that constitute vertebrates' control systems.

  Bruce Nevin

[Bill_Powers1]

[From Bill Powers (980320.1046 MST)]

Bruce Nevin (980320.1022 EST)--

... aren't you arguing on my side, that these
processes are occurring entirely within the individual? It seems to me that
you're offering a perfectly good analysis of how imitation might work,
using nothing but existing elements of HPCT.

I have no interest in using anything more than existing elements of HPCT.

There are two sorts of processes. One sort occurs entirely within the
individual (within ego), control of perceptual input. That which the
individual is observing and modelling is occurring outside individuals, ego
included.

I think there may be some misunderstanding here. Control of perceptual
input does not occur entirely within the individual: it almost always
requires acting on the external world. The perception is controlled by
means of actions that alter the state of the world outside the organism --
as an external observer perceives that world.

Behavioral outputs are a byproduct of perceptual control and so
have no status as a separate process, you say; it is an emergent
phenomenon, of no more significance than the rings and arcs observed in the
gather program.

Behavioral outputs are the _means_ of perceptual control, not a byproduct.
Without them, there would be no perceptual control (save for imagination,
which is ia different issue).

It does have status as a separate phenomenon because the individuals
involved are controlling perceptions of it as a phenomenon in their
environments.

What is "it" here? Perceptual control? Behavioral output? I don't
understand what you're saying here. Please clarify. Maybe then we can make
some progress with your other comments.

Best,

Bill P.

[bnhpct]

[From Bruce Nevin (980320.1648 EST)]

Bill Powers (980320.1046 MST)--

I'm having more than my usual difficulty being clear here. I do understand
that the control loop returns through the environment, and that behavioral
outputs are means of affecting perceptual input.

In the normal case behavioral outputs are not themselves controlled and are
not even perceived as such. You don't notice the twitches to the steering
wheel that keep the hood aligned with the roadway. What I was trying to say
is that matters of language and culture are different from "the normal
case" in this last respect. In matters of language and culture, we observe
behavioral outputs, our own and those of others, and in a secondary control
loop we control our behavioral outputs. This is not hierarchical control, a
loop at a higher level of the hierarchy adjusting the reference levels of
loops at a lower level. It is a kind of sibling recalibration, a parallel
loop in a different sensory modality monitoring behavioral outputs from the
primary loop and adjusting reference levels of the primary loop.

This applies to e.g. saying "home in" or "hone in". Control of touch and
pressure perceptions in the tongue and lips and perhaps in the alveolar
ridge results in the utterance of one or the other. This is the lowest
level of control, the primary control loop for ongoing speech.

Control of auditory perceptions can result in re-saying the words if the
first utterance is perceived as a slip of the tongue or some other error.
This is a secondary level of control, but not at a higher level of the
control hierarchy.

This second, slower loop was involved in learning the words in the first
place. It is involved in ongoing monitoring and possible adjustment of
reference levels for what the word is supposed to sound like, and can lead
to adjustment of reference levels for what it feels like to say it.

What is involved here is

o Auditory perceptions of other people speaking

o Generalizations about what kinds of people there are

o Generalizations about how the different kinds of people speak

o Control of what kind of person ego is

o Setting of reference levels for perceptions of what ego's speech should
sound like

o Setting of reference levels for tactile perceptions controlled in speaking

o Control of speaking in the tactile control loop described above

Monitoring of one's speech in the auditory control loop is the basis for
learning, for correcting errors, for keeping the tactile reference levels
calibrated, and for re-setting them depending on changing perceptions of
self image and social context.

This secondary control loop is used to re-set reference levels of another
control loop, but it is not at a higher level of the hierarchy. It is a
parallel control loop. The secondary loop involves the attempt to perceive
ourselves as we perceive other people. But we can't control those
perceptions directly, since they are perceptions of behavioral outputs. So
we adjust our control of perceptions that result in the observed behavioral
outputs, that is, we adjust the reference levels in the primary control loop.

I believe that this peculiar pairing of sibling control loops is perhaps
something new in PCT discussions. (Sibling: not hierarchical; at comparable
levels of the hierarchy, but in different sensory modalities.) I have used
language for examples, but it applies to other matters of culture as well.

Here's a restatement of something I said before to which you objected:

Observed regularities in or generalizations about behavioral outputs are
byproducts of perceptual control and so have no status as a separate
process, you say. Such regularities are an emergent phenomenon, of no more
significance than the rings and arcs observed in the gather program.

Observed regularities in and generalizations about behavioral outputs have
status as a separate phenomenon because the individuals involved are
controlling perceptions of them as phenomena in their environments and are
setting their own reference levels relative to them.

I hope this is more clear.

  Bruce Nevin

[Bill_Powers1]

[From Bill Powers (980321.0947 MST)]

Bruce Nevin (980320.1648 EST)--

In the normal case behavioral outputs are not themselves controlled and are
not even perceived as such. You don't notice the twitches to the steering
wheel that keep the hood aligned with the roadway.

There's nothing to keep you from noticing them if you want to, is there?
What you don't notice, because you can't, are the signals entering the
muscles or the contractions of the muscle fibers.

What I was trying to say
is that matters of language and culture are different from "the normal
case" in this last respect. In matters of language and culture, we observe
behavioral outputs, our own and those of others, and in a secondary control
loop we control our behavioral outputs.

You must mean "at a lower level." We observe how it feels and sounds to
speak. Specifying those feelings and sounds is the way we create
perceptions for higher systems to perceive in terms of meanings.

This is not hierarchical control, a
loop at a higher level of the hierarchy adjusting the reference levels of
loops at a lower level.

Why do you say it's not hierarchical control? We never observe our own
behavioral outputs, do we? Isn't it always some sensory consequence of a
behavioral output that we observe?

It is a kind of sibling recalibration, a parallel
loop in a different sensory modality monitoring behavioral outputs from the
primary loop and adjusting reference levels of the primary loop.

I've have to see a working model of that to believe it.

Control of auditory perceptions can result in re-saying the words if the
first utterance is perceived as a slip of the tongue or some other error.
This is a secondary level of control, but not at a higher level of the
control hierarchy.

I repeat: Why isn't it a higher level of control?

Monitoring of one's speech in the auditory control loop is the basis for
learning, for correcting errors, for keeping the tactile reference levels
calibrated, and for re-setting them depending on changing perceptions of
self image and social context.

Oh, yeah, I forgot that you think tactile/kinesthetic control is primary
for controlling speech. I think auditory control is the higher level. I
guess we'll just have to wait for data to come in.

Here's some:

Houde, John F. and Jordan, Michael I.; Sensorimotor adaptation in speech
production. Science, _279_, 20 Feb. 1998, 1213-1216.

"By use of a device that can feed back transformed speech signals in real
time, subjects were exposed to phonetically sensible, online perturbations
of their own speech patterns. It was found that speakers learn to adjust
their production of a vowel to compensate for feedback alterations that
change the voewl's perceived phonetic identity; moreover, the effect
generalizes across phonetic contexts and to different vowels." (from
abstract).

This would seem to show that the auditory control is at least important, if
not primary. To preserve the sounds that are heard, the kinesthetic details
are altered by the subject.

You might want to get in touch with the authors.

Best,

Bill P.

[bnhpct]

[From Bruce Nevin (980321.1457 EST)]

Bill Powers (980321.0947 MST)--

Houde, John F. and Jordan, Michael I.; Sensorimotor adaptation in speech
production. Science, _279_, 20 Feb. 1998, 1213-1216.

"By use of a device that can feed back transformed speech signals in real
time, subjects were exposed to phonetically sensible, online perturbations
of their own speech patterns. It was found that speakers learn to adjust
their production of a vowel to compensate for feedback alterations that
change the voewl's perceived phonetic identity; moreover, the effect
generalizes across phonetic contexts and to different vowels." (from
abstract).

This would seem to show that the auditory control is at least important, if
not primary. To preserve the sounds that are heard, the kinesthetic details
are altered by the subject.

This is very cool! It's just what I've been wishing to be able to do, and
the technical means did not seem to be available.

You might want to get in touch with the authors.

HotBot doesn't show up a site for Science (there are sites for Nature, New
Scientist, Scientific American, others). Does your source tell where Houde
and Jordan are? Local libraries don't have Science. (Falmouth stopped
taking it in 1993, which kind of surprises me with WHOI right there in
Woods Hole. I don't have access to WHOI.)

This description is consistent with the model that I have sketched.
Speakers do not immediately adjust to correct the error; they learn. I
would like to know if they gradually perturbed vowels or if they introduced
changes in such a way that control could not have been continuous.

Oh, yeah, I forgot that you think tactile/kinesthetic control is primary
for controlling speech. I think auditory control is the higher level.

I'm probably being sloppy again. I'm overemphasizing the detachment of the
higher level of control from the lower level.

I agree that auditory control is at the higher level, setting reference
levels for tactile and kinesthetic control. Here's the difference that I'm
asserting: during speaking, reference levels for tactile/kinesthetic
control are fixed. They are not varied in order to keep the sounds of the
current syllable on track during the pronunciation of those sounds.

The basis for this twofold. First, experimentally blocking auditory
feedback with no degradation of pronunciation; and second, anecdotally
recalling how anaesthesia does result in degraded pronunciation even though
there is no impairment of hearing or of physical mobility of tongue and
lips. (Or would you say that reduction of kinesthetic/tactile feedback has
the effect of overwhelming control?)

Suppose you are correct. Suppose that, in real time, in the course of
speaking, we effect our control of auditory configurations and transitions
in the sounds of our speaking by adjusting reference levels for tactile and
kinesthetic perceptions at a lower level of the hierarchy.

Consider the sound-blocking evidence. When auditory feedback is blocked, I
suppose you would have to say that the reference levels for the successive
phonemes are supplied directly to the lower-level control systems from
memory, without being adjusted in real time by the higher-level auditory
control systems. These kinesthetic/tactile reference levels are fixed. Do
you have another account for the fact that pronunciation does not
deteriorate even during an hour or so of speaking without auditory feedback?

But this is exactly what I have described. The reference levels for the
successive phonemes are fixed (at the time scale of pronouncing words),
supplied directly to the lower-level control systems from memory, without
being adjusted in real time by the higher-level auditory control systems.

I take it a bit further, supposing that this is the normal state, that our
ongoing speaking is not the control of precisely how our pronunciations of
words sound. If a variation does go so far as to be a different phoneme
sequence, we may re-say the word, and call it a slip of the tongue. That is
the extent of auditory control in normal speaking. We do not pull the sound
back within bounds for the phoneme during the course of uttering it. By the
time we hear and recognize it, it is too late. But we may readjust those
pre-set, remembered reference perceptions for the phonemes based on what we
hear. And this is my understanding of the Houde and Jordan results.

(I'm setting aside more complicated situations, such as virtuosic shifting
of dialect while speaking, talking like this kind of person and then like
that kind of person. Not germane to the immediate point.)

Here's another angle: Auditory control (by which we can reset our fixed
norms of pronunciation) is at configuration, transition, and perhaps
sequence levels. Is that too slow for controlling events like syllables and
words? What is it to learn to control a sequence as an event? Do we
establish fixed references at lower levels?

The fundamental question remains, how to test the two hypotheses.

  Bruce Nevin

[Bill_Powers1]

[From Bill Powers (980322.0121 MST)]

Bruce Nevin (980321.1457 EST)--

Houde, John F. and Jordan, Michael I.; Sensorimotor adaptation in speech
production. Science, _279_, 20 Feb. 1998, 1213-1216.

Houde's email address is houde@phy.ucsf.edu.

This is very cool! It's just what I've been wishing to be able to do, and
the technical means did not seem to be available.

You might want to get in touch with the authors.

HotBot doesn't show up a site for Science ...

Science's web page is

www.sciencemag.org

This description is consistent with the model that I have sketched.
Speakers do not immediately adjust to correct the error; they learn. I
would like to know if they gradually perturbed vowels or if they introduced
changes in such a way that control could not have been continuous.

I wouldn't call this "learning," because the adaptation is always in the
right direction. I'd rather see it just as a slow control process.

The article actually does support your position, in that the new motor
patterns persist when auditory feedback is blocked by noise (the
"adaptation" part of the experiment). I didn't see any explicit data on
how quickly the "compensation" takes place, but they did apply the
disturbance slowly over a period of 17 minutes, and subjects said they
didn't notice the disturbance or their compensating change of articulation.

Oh, yeah, I forgot that you think tactile/kinesthetic control is primary
for controlling speech. I think auditory control is the higher level.

I'm probably being sloppy again. I'm overemphasizing the detachment of the
higher level of control from the lower level.

I agree that auditory control is at the higher level, setting reference
levels for tactile and kinesthetic control. Here's the difference that I'm
asserting: during speaking, reference levels for tactile/kinesthetic
control are fixed. They are not varied in order to keep the sounds of the
current syllable on track during the pronunciation of those sounds.

I don't think you mean that tactile/kinesthetic reference levels are fixed.
If they were, you could produce only one articulation, like saying [i] and
nothing else, forever. What is relatively fixed, I would guess, is the
perceptual input function that detects the configuration of the mouth. As
the organization of this input function slowly changes, the objective mouth
configuration corresponding to a given articulation reference signal
changes. This is the long-term adaptation. Short-term, however, one can
still rapidly alter the reference signal for which articulation is to be
produced, say along the scale from "eee" to "ooo", and the actual sensed
articulation will immediately follow the reference signal. So the auditory
feedback effect is very rapid, as rapid as one's ability to say consecutive
phonemes. The articulatory control must operate as rapidly as speech can
proceed.

If you do contact the authors, you might want to ask them how long the
compensation takes. Since they used a very slow disturbance, they might not
know, but they could certainly do a quick test to see how rapidly a
step-disturbance would be compensated. They used the analogy of the prism
goggles, and we know that in that case, compensation is very rapid,
although adaptation takes a long time. If the speech system works in a
similar way, we would expect a person to correct the heard sound quickly if
the initial articulation was in error. After all, if a person said "peg"
and heard 'pig", there would be a large auditory error, and one would
expect the reference signal to be shifted quickly toward the articulation
for "peeg" in order to get the sound of "peg." This would apply to all
words, of course (what the authors call "generalizing").

In the experiment, subjects were asked to read from a list of words for a
very long time, to get the adaptation effect. By the end of this time, I
infer that speech was proceeding at a normal speed with the subjects
hearing the correct pronunciation but producing a shifted set of
articulations. I don't know what the effect would be if the subjects were
reading from continuous text or speaking spontaneously. Small disturbances
might be so quickly compensated that subjects wouldn't realize they were
articulating any differently. This is something to ask Houde about.

The adaptation could actually be in the output function of the auditory
level of control. After all, the disturbance is not mechanical, but
acoustical. One would have to think carefully about just where the
adaptation must be taking place. The disturbance is not like a mechanical
one that changes the dependence of the acoustical cavity shapes on muscle
lengths (like Demosthenes' stone in the mouth). This will require some more
thought.

The basis for this twofold. First, experimentally blocking auditory
feedback with no degradation of pronunciation;

This certainly tells us something, but it's like Blom's "walking in the
dark" example. People may use auditory feedback when it's available, and
switch to kinesthetic control when it's not. Before I will accept "no
degradation of pronunciation" I would would want to see sonograms: human
listeners can tolerate large changes in auditory inputs.

and second, anecdotally
recalling how anaesthesia does result in degraded pronunciation even though
there is no impairment of hearing or of physical mobility of tongue and
lips. (Or would you say that reduction of kinesthetic/tactile feedback has
the effect of overwhelming control?)

The latter -- loop gain is determine by _all_ sensitivity factors around
the loop, multiplied together. However, all these perturbations are
legitimate ways to sort out where in the loop various effects are occurring.

Suppose you are correct. Suppose that, in real time, in the course of
speaking, we effect our control of auditory configurations and transitions
in the sounds of our speaking by adjusting reference levels for tactile and
kinesthetic perceptions at a lower level of the hierarchy.

Consider the sound-blocking evidence. When auditory feedback is blocked, I
suppose you would have to say that the reference levels for the successive
phonemes are supplied directly to the lower-level control systems from
memory, without being adjusted in real time by the higher-level auditory
control systems. These kinesthetic/tactile reference levels are fixed. Do
you have another account for the fact that pronunciation does not
deteriorate even during an hour or so of speaking without auditory feedback?

This is not an easy problem. The biggest problem is why, when auditory
feedback is swamped by noise, the outputs don't wildly exaggerate the
reference signal changes for the articulatory systems. This is what
inclines me to accept your idea of a purely kinesthetic channel (at several
levels) with the acoustic systems being used to trim the calibration of the
kinesthetic control systems. This is where Houde's method will be
invaluable. If it should prove that a step-disturbance is followed by a
very slow compensation that takes place only after many words have been
spoken, your basic concept would be clearly supported, and my idea of rapid
auditory control would be ruled out.

A clear answer to this question seems within reach, with Houde's apparatus.

The fundamental question remains, how to test the two hypotheses.

I think it can be done, with Houde's cooperation.

Best,

Bill P.