Alerting

[From Rick Marken (940609.1230)]

Martin:

In the matter of alerting systems, PCT shows the necessity, not the
mechanism.

Me:

I think Martin actually believes that a theory (PCT) can show the
"necessity" of an explanatory mechanism

Martin Taylor (940608 18:40) --

I'm not sure what you mean about a theory showing the necessity of an
explanatory mechanism.

I was just echoing what you said, above. I read it as "PCT shows the
necessity of some kinds of alerting system mechanism, though PCT doesn't
show what the particular mechanism is". Did I derive the wrong meaning from
your sentence?

A thoery IS an explanatory mechanism, and as I see it, a theory that showed
the necessity of an explanatory mechanism would do so only by demonstrating
its own inadequacy in explaining some phenomenon within its domain.

Yes. I agree. So PCT was unable to explain some data (phenomenon) and,
therefore, you were led to see the necessity of an alerting system mechanism?
What data did PCT fail to explain? What did the PCT model that failed look
like?

I think I can say that my statement could be rephrased more accurately as
"If PCT is true, then necessarily there must be an alerting system," or,
"conditionally on PCT being true, then it shows the necessity of alerting
systems."

It still seems like you are deriving the need for an explanation of a
phenomenon from the explanation for (possibly) another phenomenon. I think I
understand PCT pretty well and I just don't undertand how the theory shows
the necessity of an alerting system.

There are observations such as that one's attention tends to be attracted
by changes in the visual periphery in patterns one cannot statically
discriminate, or that although one can focus one's auditory attention
nevertheless there are sounds that tend to redirect it, or that no matter
how engrossed one is in typing a response to CSG-L, a touch on the neck
will divert that attention when sound or sight of an intruder might not.

But all these things can be explained by PCT without any new mechanism; your
own dead-zone proposal might work, as well as Bill's hierarchical approach
(higher level conflict resolving system). I think many of these observations
can be explained in terms of the non-linear error function that works so
nicely at "capturing" a perception once it gets into a certain range of the
reference state but ignoring it otherwise. It looks like the perception
"alerts" the control system; but it's just a regular, input control system,
with an error (comparator) function that looks like this: __/\ __.
                                                              \/

Why does PCT show the need for some kind of attention-shifting system based
on events (I don't mean the event level) in perceptions not under control?

That is the question.

Your answer, in terms of degrees of freedom, does not move me. I would rather
see just one observation of an altering phenomenon -- some DATA -- that
requires some kind of alerting mechanism.

Given that there is such a large excess of controllable perceptions over
perceptions being controlled, there MUST be some way that control is
shifted from one to another AS THE NEED ARISES.

Just demonstrat one instance where this is the case. Your analysis seems to
imply observations that could be made to show that "there MUST be a way that
control is shifted from one to another AS THE NEED ARISES". Otherwise we're
in "angels on pins land", which is even more like LA LA land than LA itself.

Is it clear now how PCT explains the phenomenon of alerting systems?

I already knew how PCT could explain the phenomenon of alerting. I just
haven't seen any quantitative "alerting" data that could be used to test a
model.

Alerting systems, by the way, are not a phenomenon to be explained -- at
least as I understand the term. An alterting system (as I understand it) is a
POSSIBLY necessary explanatory mechanism; it explains an as yet unspecificed
(quantitatively) phenomenon called alerting. The anecdotal descriptions of
alerting mentioned in your post can all be handled (anecdotally) with no
change or addition to the PCT model. Quantitative data is needed to see if
this is quantitatively true.

So I guess the answer to your last question is "no".

Best

Rick

[Martin Taylor 940610 10:40]

Rick Marken (940609.1230)

A thoery IS an explanatory mechanism, and as I see it, a theory that showed
the necessity of an explanatory mechanism would do so only by demonstrating
its own inadequacy in explaining some phenomenon within its domain.

Yes. I agree. So PCT was unable to explain some data (phenomenon) and,
therefore, you were led to see the necessity of an alerting system mechanism?

Wrong. PCT WAS able to explain a phenomenon that previously stood on its
own as something special. That's why I brought it up. The existence of
the alerting mechanism was something I wrote about as a model for how we
should use the information from our remote sensing satellites, in 1972.
I don't know how long before that the function was known about--probably
Aristotle, or William James knew about it.

What data did PCT fail to explain? What did the PCT model that failed look
like?

None. That's what is so nice about it. (By the way, I think you should
more carefully distinguish PCT the theory from PCT the model using only
the most trivial kind of control system. Nothing in PCT the theory dictates
what kind of control system is used at any particular level of the hierarchy.
This comment is triggered by the present discussion, but doesn't refer only
to it).

The anecdotal descriptions of
alerting mentioned in your post can all be handled (anecdotally) with no
change or addition to the PCT model.

Yes, sure. I don't see what all the fuss is about. Why do you think I
want to change the model (let alone the theory)? I only require there
to be an alerting mechanism. Plausible mechanisms that require no change
to the theory have been proposed. If they work, they work. So what?

For a long time, there were physiological and psychological observations,
standing on their own as things to be explained. PCT explained why they
are as they are, and when I point this out, I am treated as some sort of
heretic for daring to point out the power of PCT in an area that hadn't
been dealt with in BCP.

At least that's the way it seems to me. Sorry if I disturbed you by pointing
out an unnoticed aspect of the power of PCT.

Quantitative data is needed to see if this is quantitatively true.

I provided some, but you say

Your answer, in terms of degrees of freedom, does not move me.

I don't know what else would be satisfactory. I suppose a "one-armed
paper-hanger" experiment might do, where the subject is working to capacity
on some task, and there is another task needing attention. The times
when the other task needs attention could be signalled by providing input
to sensory signals needing alerts. But we do this all the time in everyday
life, so I don't know what it would show. And at least in HCI there is
a good demonstration of it. A simulated interface to a soft-dring bottling
plant with two operators was constructed (I'm blocking on the name of the
researcher for the moment). Every function of the plant was available on
the screens of the two operators, and their job was to keep it running
correctly in the face of simulated malfunctions such as blocked valves,
excess or insufficient supply of bottles, etc. In one condition, they
could see on the screen everything that was happening, and in the other
they had a kind of acoustic representation as well, with glugging, chinking,
clattering, humming and other sounds. Performance was MUCH better in the
second case, in which subtle changes in the sound pattern alerted the
operators that something was amiss (without necessarily telling them
what was wrong).

I suppose that this isn't alerting either.

Oh well. I'm used to it.

Martin

[From Rick Marken (940610.1500)]

Martin Taylor (940610 10:40) --

PCT WAS able to explain a phenomenon that previously stood on its
own as something special.

What phenomenon? Where's the data? Behavioral data.

That's why I brought it up. The existence of the alerting mechanism was
something I wrote about as a model for how we should use the information
from our remote sensing satellites, in 1972.

The "existence" of the alerting mechanism implies that "alerting mechanism"
is a phenomenon. Why call it a "mechianism". This is getting very confusing.
What's the "alerting" phenomenon?

I said:

Quantitative data is needed to see if this is quantitatively true.

You say:

I provided some

No. You provided a bunch of words. By data, I mean the results of
experimental studies using real, soft, cuddly organisms.

I don't know what else would be satisfactory.

Experimental data. You've heard of it, I'm sure ;-).

I suppose a "one-armed paper-hanger" experiment might do

If it provides real data, yes indeed. But there should be experiments in the
literature; or perhaps easy demos that we could do right in our own homes,
without fancy equipment.

Performance was MUCH better in the second case, in which subtle changes in
the sound pattern alerted the operators that something was amiss (without
necessarily telling them what was wrong).

I suppose that this isn't alerting either.

If that's what you're talking about, great. It's "alerting". And a tap on the
patellar tendon is a "stimulus". And a food pellet is a "reinforcer". And a
chess move is a "programmed output" PCT, as is, shows that it's all just
control. Talk of "alerting mechanisms" as an explanation of this phenomenon
is as incorrect as talk of "unconditioned responses" as an explanation of
the knee jerk or (as Bill Powers (940609.0910 MDT) said) of "sucking" as an
explanation of how a vaccum works. If the alerting phenomenon is what you
describe, then it is simply "control".

I can model, using a simple control system, the phenomenon you describe as
"alerting". So there is no need for a special alerting mechanism and the
phenomenon itself is just another view of control (it's what happens when
there is a sudden disturbance to a variable that has been remaining -- up
to that point -- in it's reference state without the need for any action).
"Alerting" now takes its place along with "reinforcement", "stimulus" and
"programmed output" as a misinterpretation of a control phenomenon. It could
be added as the fourth blind man in the "Blind men and the elephant" paper.
It is no more necessary now than "epicycles" were after Newton.

Best

Rick

[Martin Taylor 940613 11:00]

Rick Marken (940610.1500)

If the alerting phenomenon is what you describe, then it is simply "control".

Talk of "alerting mechanisms" as an explanation of this phenomenon
is as incorrect as talk of "unconditioned responses" as an explanation of
the knee jerk or (as Bill Powers (940609.0910 MDT) said) of "sucking" as an
explanation of how a vaccum works.

Did I ever say it was an "explanation?" I THINK I said PCT PROVIDED the
explanation. I guess you just don't like "new" words, or the idea that
PCT explains things you hadn't noticed that it explained.

I can model, using a simple control system, the phenomenon you describe as
"alerting".

If you are taking normal English usage, "a" implies the singular. And if
that's what you mean, you aren't talking about the same situation at all.
A hierarchic (and therefore not "simple") control system may well behave
as specified.

The main point is that, as Bob Clark keeps saying in other contexts, there
must be ways in which control shifts from one perceptual signal to another.
Clearly the immediate mechanism involves changes in higher-level reference
signals. But as with "simple" control, changes in output correlate with
the effect of the disturbance. So with alerting, changes in WHAT is
controlled relate to changes in disturbances. And different senses (in
humans) seem to have different probabilities of affecting those changes.
(And note the spelling: I did NOT write "effecting.")

Martin