[From Erling Jorgensen (2005.01.15 0230 EST)]
Richard Kennaway (Fri, 13 Jan 2006 16:18:09 +0000)
Your comments lead me to some reflections, not so much
about the "mirror neurons" research, as to the core
PCT model. (BTW, I know you know most of this, but there
are potentially 99 of us involved in this conversation, at
one remove or another. What one person might not find
intriguing, another might...)
It may be interesting that there is such a localisation
of specific (classes of) perception to specific neurons,
but the popular extrapolations about how this explains
empathy seem to me unwarranted. What am I missing?
You're catching one of the main problems with these
reports, as far as I can tell. To claim that one is
explaining empathy is to run off enamored with possible
conclusions, before the phenomenon has even been nailed
down.
But let's go back to the first part of your sentence --
It may be interesting that there is such a localisation
of specific (classes of) perception to specific neurons,
This is of interest, I believe, from the standpoint of
the PCT model. In 1973, Bill Powers proposed a functional
model of how to apply the insights of cybernetics --
specifically, negative feedback control -- to psychological
phenomena. It was a retooling of Norbert Wiener's
work, but specified from the point of view of the control
system, not the external observer/designer/engineer.
An important part of Powers' model was its modularity.
It was built out of elemental control loops, with a
hierarchical proposal for how to link them together.
Subsequent working demos and simulations, by Powers
and others, confirmed that a hierarchical arrangement,
properly tuned, could produce stable control of multiple
variables simultaneously, without unstable oscillation.
This body of work constituted a _prediction_ about the
functional relationships that one might (must?) find,
in how living systems, specifically human, operate.
Powers attempted to be as parsimonious as possible with
his model -- as is true of all good models -- keeping to
the minimum number of concepts necessary, to try to
account for human functioning. And here I refer simply
to the operationalized components of the basic control
system: namely, two variables within the loop, perception
and output, two variables from outside the loop,
reference and disturbance, and a couple of parameters,
slowing factor and gain -- the latter of which could be
partitioned into various points around the loop, but
which for convenience is generally included in the
output equation. There was a further principle, that
if basic control systems were linked hierarchically,
then each higher level _had to_ operate with a slower
time constant.
That's it. That's the basic model. This was a functional
description, entirely specified as to its internal
relationships, which appeared to apply to a growing
range of perceptions-under-control, as later simulations
began to demonstrate.
What was not pre-specified was the specific implementation
of these functional components within the (human) nervous
system. Essentially it was a prediction, that _no matter
how_ a perception might actually be constructed, and _no
matter how_ a motor output might actually be constructed,
these functional relationships should hold! That was an
amazing prediction.
As a secondary elaboration of his basic model, Bill Powers
made some educated guesses about the classes of perception
that one might find humans operating with. This is the H
in Hierarchical PCT, with at present 11 qualitative types
of perception, seemingly arranged in some kind of layering.
This has never been an essential part of PCT. For myself,
I treat it as a "grounded qualitative theory," in the
technical sense of that term, with very useful heuristic
properties.
Nevertheless, these hierarchical layers of perception
were also _predictions_ -- not essential to the operation
of the model, but proferred nonetheless. Powers was
saying we should be able to find something in human
neurophysiology corresponding to classes such as these.
Enter, the "mirror neurons." In light of this second
tier of predictions made by Powers, the fact that these
researchers found (to use your words) "such a localisation
of specific (classes of) perceptions to specific neurons",
is indeed interesting. To call them "cells that read
minds" certainly seems unwarranted. But to call them
"perceptions caught in the act of perceiving" may be
another matter, of no small significance.
What surprised the original researchers, as far as I can
tell, was that "observing" and "acting" should utilize
the same sets of neurons. This seems to be because
observing and acting -- according to their (implicit)
model of how to categorize phenomena -- appear to be
such different behaviors. But a model that controls
perceptions by closing the loop between what is acted
upon and what is observed, does not have that same
problem.
What was anomaly to them, does not seem so to us on
CSGNet. This seems to be the import of your question --
"Is there less here than meets the eye?" We ask, "What
was the monkey perceiving that could have been the same
in either condition?" And we have put forward a few
candidates, such as the perception of "grasping" itself,
or that of a transition (and/or relationship) such as
"toward-the-mouth," or perhaps the event-perception
(or is it a program) of "eating."
The researchers observed an apparent regularity in the
firing of the (so-called) mirror neurons. To a PCT way
of thinking, we see stabilized regularities all the time
-- we either call them controlled perceptions, or the
reference standards to which controlled perceptions are
brought.
So the anomaly of this research is a little different
for those of us using the PCT model. We already know
how observing and acting can be implicated in a common
control loop. What is unclear to me is whether passive
observation is indeed control.
See, that's the interesting question that I think this
research suggests. It seems the mirror neurons were
already found to be predictably operative in certain
actions of the monkey (eating a peanut, or whatever.)
According to our PCT model (and slogan!) of such things,
all behavior is the control of perception. And even
the varying behavior that is part of controling a higher
level perception, is itself enacted as a control process,
akin to pursuit tracking.
So then, if the same neurons are firing in a (sufficiently)
same way with both active control and passive observation,
it seems that both are instances of control at a similar
level.
I can see how continuing to observe someone else eat a
peanut (or whatever) entails some form of control. There
is at least one stabilized variable apparent, which is
maintained despite disturbances -- the very act of
"continuing" with the observing. But as Bjorn Simonson
(2006.01.13,11:45 EUST) has pointed out, the _nature_ of
the observations are quite different when one is doing the
eating oneself. Tactile monitoring comes into play, and
orientation is vastly different, and the outcome leads
to tasting and actually eating. So, how could the control
be precisely at the same level? ...(so much so that the
locale of cell firing is apparently the same.)
Assuming we get past the question of which perception
at which level is jointly involved, I think a deeper
issue is raised by this matter of observation. Control,
by definition, means matching-to-template. When we
simply observe something, are we matching it to a
template? The notion of "recognition" seems to contain
at least an implicit sense of template in its etymology,
as a "re-cognizing."
Is there a form of observing that foregoes all templates?
There are certainly all kinds of background sensations
that we routinely disregard, and in that sense we are
not controling for them. But when we shift and attend to
them, aren't we then actively controling the perception
of keeping them in attentional focus. And don't we have
a colloquial term for this, which incidently uses a
purposive word? -- we call it "listening intently."
Was the monkey with the "mirror neurons" observing (the
peanut!) intently? Or was it just setting a reference
for wanting that peanut? -- the same sustained reference
it used when it ate peanuts itself. Maybe, in the case
of these mirror neurons, it is not so much perceptions,
but rather _a reference signal_ "caught in the very act
of wanting."
Mirror cells do not detect intentions. They enact them!
All the best,
Erling
NOTICE: This e-mail communication (including any attachments) is CONFIDENTIAL and the materials contained herein are PRIVILEGED and intended only for disclosure to or use by the person(s) listed above. If you are neither the intended recipient(s), nor a person responsible for the delivery of this communication to the intended recipient(s), you are hereby notified that any retention, dissemination, distribution or copying of this communication is strictly prohibited. If you have received this communication in error, please notify me immediately by using the "reply" feature or by calling me at the number listed above, and then immediately delete this message and all attachments from your computer. Thank you.
<<<<RCMH>>>>
