Attention (was Re: Sequence and ...)

[Martin Taylor 2019.]

Attention, whatever else it is, is a means of making conscious some

perceptions at the expense of others. Attended perceptions need not
be based on current sensory data. They may be entirely in
imagination, as seems to occur when someone “lost in reverie” (or a
scientist trying to resolve some mathematical or theoretical
problem) doesn’t seem to hear a call for dinner. In that respect, I use “seem to hear” rather than “hear” not only
from scientific caution, but because of a long ago experience, when
I was called to join a friend for lunch. My conscious experience was
that I immediately got up to join him, but when I went out into the
corridor, he was nowhere to be seen. I went quickly to the
cafeteria, where I found him already in the queue. This friend
walked very slowly with a cane, and must have taken several minutes
to reach the queue down two long corridors, but I heard his voice
those several minutes after he spoke, having been absorbed in what I
was (mentally) doing.


On 2019/07/12 12:19 PM, Bruce Nevin
( via csgnet Mailing List) wrote:

      [Bruce Nevin

20190712.12:05 ET]


Taylor 2019.Â --Â

          the Program

level control was done consciously. Conscious control is
much slower than control of the same perception when it
has become built into the hierarchy Â

      This is what

I meant by training and skill.

      This is an

important aspect of learning: what is it that happens while
control of a perception becomes ‘built into the hierarchy’?
Previously, attention was required in order to control that
perception; subsequently, attention is available for other
purposes, perhaps not limited to the purposes for which that
perception is being controlled?


‘attention’ is, it is (always?) circumscribed by limits on
means of perception and means of effecting control. Eyes
directed over there  cannot be simultaneously directed
over here , similarly for orientation of torso and
limbs, orientation of head/ears for hearing and (I think
poorly understood) control processes for attending to some
sounds and ignoring others, etc. Whatever attention is
–perhaps it is a function of problem-solving systems; if so
those systems are constrained by such limitations on their
means of perceiving and of effecting control, whence
limitations on the field of attention.Â


      On Thu, Jul 11, 2019 at 5:24

PM Martin Taylor < >

[Martin Taylor 2019.]


don’t think there’s much advantage either to the
protagonists or the onlookers to rerunning the argument to
which Bruce referred, which was generated by your
Sequence-Program demo. We may have differing reasons why
your demo generated a fairly long confusion, but I don’t
think there’s any difference of opinion about the fact
that it did.

          So I intend to bring up a point that I don't think was

mentioned earlier – that the subject in the demo did not
have time to build into the hierarchy the perceptions and
actions in the demo for the “Program” perception. Maybe
they did for the Sequence component of the demo, but it is
clear that the Program level control was done consciously.
Conscious control is much slower than control of the same
perception when it has become built into the hierarchy, as
many examples make clear.

          Yes, it is necessarily correct that control of a

higher-level perception is slower than control of a
contributing lower-level perception. You don’t need
experiments to demonstrate that fact. If a perception at
level N is an input to a perceptual function at level N+1,
the level N+1 perception can’t be well controlled while
the level N perception is still finding its reference

          As a separate point, the multiplex perceptions and the

actions required in the demo are sufficiently similar to
those of a Stroop Test to make me wonder whether some of
the same slowing effects might be operative in both tasks.

          Bruce has pointed out that your "*                    if the shape is

circle, the next color is blue; else, the next color
is red* " can be represented
as a set of overlapping two-element sequences. Those
conflicting overlaps have much in common with a Stroop

              You are right about the imprecision of language to

describe what is going on, but that doesn’t affect the
fact that there are different possibilities about what
is actually going on, just as in the
coin game the similarity of forms represented by “Z”,
“N”, and “Zig-Zag” allows different descriptions of
the same thing on which subject and experimenter
agree, nevertheless they may differ perceptually to
the subject in such a way that one more replacement of
a coin would lead to the subject saying “No” for N and
“Yes” for Z. The verbal agreement would become a
verbal disagreement, because the subject was not
controlling what the experimenter had thought.


[Rick Marken 2019-07-11_11:20:19]

                  [Martin Taylor



2019/07/8 2:27 PM, Bruce Nevin (
via csgnet Mailing List) wrote:


Nevin 20190708.14:25 ET]Â Â

                                  BN: Completely out of

awareness is the level from which
we control the level that we are
observing. This has confused
discussions of how to demonstrate
control at a Program level, and
the distinction between Program
perceptions and Sequence

                  MT: That may be the reason, but I think the

confusion comes from quite a different source, the
inherent difficulty of determining whether a
perceived sequence is a Sequence perception only,
or is a fragment of a Program perception.

                RM:Â  It seems to me that what confused the

discussion of “how to demonstrate control at a
Program level, and the distinction between Program
perceptions and Sequence perceptions” was similar to
what confused the results of the Test for the
Controlled Variable in Bill’s description of the
“coin game” in B:CP (pp. 236-238 in the second
edition). As Bill says in that section: “Another
educational feature of the game is how it * puts
verbal analysis aside.* I know of no clearer
demonstration of the difference between perceiving
and talking about perceptions. When E and S compare
written definitions at the end of the game, they may
often find that they have used quite differentÂ
language, different verbal analogs of the controlled
quantity. S may have been preserving a “zigzag”
pattern, and E may conclude it is the letter N or Z.
If they are both word-oriented types, E and S may
argue about whose definition is the “right” one,
forgetting that E has discovered what S was in fact
controlling, whatever either of them likes to call

                RM: The "confusion" in the  Program control

discussion seemed to be precisely of this kind; it
turned on a difference in the way people like to
talk about what was being controlled while ignoring
the fact that the program control demo ( )
reveals what people are actually controlling,
regardless of what you want to call it. Bruce likes
to call the program that is controlled in that demo
a sequence; Martin likes to call the program a
“branching network in which the links are sequences
of states or
events”. The implication is that the the program
control demo doesn’t actually demonstrate control
of a program perception. But this ignores the fact
that the demo shows that what I call a “sequence”
perception (small, medium, large) can be
controlled at a much faster rate than what I call
a “program” perception ( if
the shape is circle, the next color is blue;
else, the next color is red). Â

                    RM: This finding is consistent

with the idea that different types of perceptions are controlled
at different levels of the hierarchy, and that
perceptions lower in the hierarchy can be
controlled at a faster rate than those higher in
the hierarchy. And there is quite a bit of
evidence that this is the case.  So I take
the results of my program control demo as evidence
that what I call “sequences”, such as " small,
medium, large",  and “programs”,
like " if
the shape is circle, the next color is blue", are two different
types of perceptual variable that are
controlled at two different levels of the control
hierarchy. This conclusion may be wrong, but
whether or not it is or not cannot be determined
by verbal argument. It has to be tested, using the
method used in the program control demo, that
method being a version of the test for the
controlled variable.Â




                  Programs can be seen as a

branching network in which the links are sequences
of states or events, but in the execution of any
one of them, all you ever see is a single

                  Let's consider a hypothetical perceived sequence

“red-blue-green”. should this be described in
words as

                  (a) red, blue, green.

                  (b) If red then blue;

                  Â Â Â Â  if blue then green;

                  Â Â Â Â  if green then stop [or perhaps "if green than


                  (c) If red then blue

                  Â Â Â Â Â Â Â  else

                  Â Â Â Â Â Â Â Â Â Â  if blue then green

                  Â Â Â Â Â Â Â Â Â Â  else

                  Â Â Â Â Â Â Â Â Â Â Â Â Â  if green then stop [or "if green the


                  Â Â Â Â Â Â Â Â Â Â Â Â Â  else

                  Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â  report error;

                  Â Â Â  endif;

                  These possibilities imply different levels of

perception. They are reference values for (a) a
sequence level perception, and (b) and (c)
different complexities of Program level
perceptions. If you perceive red-green-blue or
yellow-black-white, there would be an error in
that control loop. But what action might correct
the error? Would it differ depending on whether
the controlled perception was a Sequence or a
Program? How would you perform a TCV to determine
which level was the controlled perception?

                  Surely Sequence and Program are different as

reference values. But how would you distinguish
them experimentally? It was this question, applied
to Rick’s demo, that generated the confusion in
the first place, so I think that if there is an
experimental way to distinguish them. something
more complex than that demo would be required.

                  Maybe what should be distinguished as a level is

not Program from Sequence, but “Choice-point” from
control of magnitude. Or maybe there is a
definitive way to distinguish program control from
sequence control.


Richard S. MarkenÂ


is achieved not when you
have nothing more to add,
but when you
nothing left to take away.�
 --Antoine de Saint-Exupery